brain initiative notes - biafra ahanonu Notes on neuroscience related news.. 2015-11-22T06:41:44+01:00 2015.11.19 biafra ahanonu 2015-11-21T00:00:00+01:00 2015-11-21T00:00:00+01:00

2015.11.19 [link]

New voltage sensor: Ace2N and Ace1Q


Figure 1: (right) Ace2N voltage sensor construct and example cells. (left) Example optical (top) and eletrical (bottom) trace from the same cell.

A new paper from my lab has been released chracterizing Ace2N, a new voltage sensor.

go to post.

2015.10.30 biafra ahanonu 2015-11-21T00:00:00+01:00 2015-11-21T00:00:00+01:00

2015.10.30 [link]

Society for Neuroscience 2015 notes: pain and reward circuitry

Note: I’m breaking up my notes from SfN into a series of posts focusing on a specific topic and mainly giving summaries of interesting posters, talks, etc. I came across.

go to post.

2015.10.22 biafra ahanonu 2015-11-21T00:00:00+01:00 2015-11-21T00:00:00+01:00

2015.10.22 [link]

Society for Neuroscience 2015 notes: pain technologies

Note: I’m breaking up my notes from SfN into a series of posts focusing on a specific topic and mainly giving summaries of interesting posters, talks, etc. I came across.

There were a lot of new pain-related technologies on display this SfN, mostly with a focus on


Figure 2: Epidural optic fiber implant for spinal optogenetics

* BB9 - Epidural optic fiber implant for spinal optogenetics

* They are able to do inhibition in the spinal cord, but have mentioned that this is likely to be challenging overall.

  • DD3 - Optogenetic control of dopamine release in rodents and novel opto-dopamine probes for In vivo experiments
    • Can stimulate DA terminals and do fast scan cyclic voltammetry at the same time.
    • Takes about 1 minute for DA to recover its responsiveness, this is tested with varying intervals between light stimulation.
    • Tail pinch induces maybe a slight decrease in DA release in NAcc but see a weird oscillation in activity for several tens of seconds starting about 10 seconds after the tail pinch stops. Asked if he’s tried other pain stimuli, but didn’t indicate that they had.
  • 549.02 - Optogenetic control of pain and motor circuitry
    • Outlined several designed traits in peripheral optogenetics
      • Expression over long periods of time
      • Dealing with increased movement, e.g. flexing of the spinal cord
      • Not immune privileged, e.g. don’t have BBB
      • Tissue opacity, especially in the spinal cord
    • Halorhodopsin had decreased sensitivity when placed in DRGs and need constant light illumination which might cause heating and other problems
    • To combat this, used SwiChR, which caused a 0.4 to 1.3g withthrawal threshold increase when used. Also tried iC1C2, NpHR, and YFP with no effect.
    • Increasing SST+ interneurons is aversive (at least with cFos proxy).
    • Inhibiting SST+ with hM4Di causes an increase in thresholds
    • They have tried multiple fibers along the spinal cord for optogenetics, this did not work.
    • Temperature effects on optogenetics is important.
    • AAV6-hSyn-iC++-YFP might be a useful tool for better inhibition.
    • Inhibition of muscles is possible
      • They tried a number of other variats of AAV and inhibitory channels/pumps to no effect.
    • Cannulation of mice does not affect pain thresholds, at least for Amy’s prep.
    • Papers
  • 549.03 - Optical control of stem cell derived motor neurons restores function to paralysed muscles

  • PIC
    Figure 3: Anna Poon wireless delivery to mice

  • 549.04 - New paradigms in wireless light delivery
    • See recently published work:
    • Several others working on wireless optogenetics
      • Hirase, Riken
      • Boyden, MIT
      • Roger, U of Illinois
    • Wireless device requirements that they wanted to meet
      • Don’t need to handle animals
      • Long form experiments
      • Small device
    • Mouse as a dielectric resonator, so that the mouse enables focusing of RF field and don’t need complicated tracking. They didn’t test on larger animals, but same principle should work. She noted that small animals are dielectrical objects that support specific electromagnetic modes.
    • There is a small amount of heating (about 1C) depending on the duty cycle of the signal generator.
    • They are able to get place aversion with the wireless device and can put multiple animals together.
    • They are NOT individually addressable at the moment, they are building a silicon chip design that might allow this.
    • Multi-color and behavior triggering of the light should be allowed via a small silicon chip.
    • Designs for the devices have been made available.
    • The whole setup is silent.
    • With rats, higher order modes are excited when box is made larger, so would need to adjust.

  • PIC
    Figure 4: An Optogenetic Demonstration of Motor Primitives in the Mouse Spinal Cord

  • 549.05 - An Optogenetic Demonstration of Motor Primitives in the Mouse Spinal Cord
    • Looking at muscle coordination and motor primitives.
    • Modular muscle stimulation allows probing of how different parts of spinal cord affect forces produced by muscles.
    • Showed that convergent and parallel isometric force fields are located at the dorsal and ventral spinal cord, respectively.
    • See
    • Mapped leg responses in rodent by light stimulation in Thy1-ChR2 mouse line
    • Created a system that randomly moves the mouse’s leg around, stimulates spinal cord, and measures the direction of force. Doing this over an entire area allows one to create force maps.
    • Motor neurons have ˜9ms delay and produce convergent fields.
    • ChAT excitatory interneurons produce parallel fields with ˜6ms delay.
    • Thy1 force fields have a wider distribution that ChAT neurons.
    • MN and ChAT force fields are additive, e.g. there is a linear superposition of parallel and convergent fields.

  • PIC
    Figure 5: Optogenetic control of aversive sensory circuitry

  • 549.06 - Optogenetic control of aversive sensory circuitry
    • How is itch coded? Discussed the neural basis of itch.
    • Histamine is classically thought to be a pure sensation of itch but when placed in the deep muscle, causes pain.
    • Capsaicin is thought to be for pain, but when topical capsaicin is placed on the skin, causes itching.
    • Assumption that there is some lateral interaction that helps sharpen sensory acuity.
    • Looks at B5-I spinal interneurons neuron induced itch.
    • Inhibition of B5-I neurons stops itching behavior.
    • They noticed that B5-I activation seems to silence a random interneuron.
    • Ex vivo skin prep, see
      • apply cutaneous stimulation to the skin (including cowage, itching powder)
      • modulate interneurons in the spinal cord via ChR2
      • record from projection neurons via patch
    • 5HT can also be used as a puritagin
    • B5-I neurons inhibit itch via feedfoward inhibition
    • KOR agonist inhibits itch as well, pointing toward opioid release from B5-I neuron
    • Some B5-I neurons seem to directly inhibit PNs
    • Noticed that it is easiest to trigger APs from terminal endings
    • Light delivery isn’t the same as natural stimulation
      • Cfiber, get 4Hz stimulation response, but with opto it is variable around 2Hz.
      • Afiber, get 6Hz natural stimulation response, but with opto it is reliably at 2Hz.
    • Keratinocytes are sufficient to trigger APs if activated, can block stimuli responses by inhibiting keratinocytes.
    • Papers        
  • 549.07 - Optogenetic dissection of visceral pain
    • Bladder pain syndrome is normally treated with short duration anesthetics.
    • Continuous vs. pulsed input causes different behavior.
    • Optogenetic modulation of distension can induce bladder pain.
    • eArchT activation in sensory neurons can inhibit bladder pain responses when bladder is artificially swelled.
    • Developing wireless LED based on previous technology
    • RF powered LED device, see Neurolux:
      • ([]
    • How stable is RF field? Will this fluctuate and how does it compare to Ada Poon’s system?
      • 20 cm up from the bottom of the cage leads to only a 10% reduction in power
      • Elicits spontaneous pain and real time place preference.
  • Q3 - Characterization of optogenetic activation of non-peptidergic C-fibers
    • See Ross minisymposium
    • Using ex vivo skin prep to characterize PN responses to MrgD neuron stimulation.
    • Need 2Hz, but not 0.1Hz, stimulation to induce Mrgd-Cre response to blue light stimulation.
    • Lamina I PN show responses to multple cold/hot, mechanical, and light stimulus protocols.
    • No hyperreactivity of PNs after inflammation when stimulating MrgD neurons.

go to post.

2015.10.11 biafra ahanonu 2015-11-21T00:00:00+01:00 2015-11-21T00:00:00+01:00

2015.10.11 [link]

dual photo-stimulation and imaging, review


Figure 6: Overview of current technologies and protein systems used for recording from and manipulating neurons.

I’ve written several times about dual photo-stimulation and imaging (dual photo-stimulation and imaging, freely moving dual photostimulation and imaging, dual photo-stimulation and imaging, cont’d). It looks like this SfN will have a lot of posters, talks, and excitement about this new technology. A new review in Journal of Neuroscience by several of the leading investigators in this area (Emiliani, Cohen, Deisseroth, and Husser) gives a nice overview of several recent technologies in the area and future directions.

go to post.

2015.09.27 biafra ahanonu 2015-09-27T00:00:00+02:00 2015-09-27T00:00:00+02:00

2015.09.27 [link]

Science is hard

Haven’t uploaded previous posts in awhile, I’ll be adding the backlog of notes and papers in the coming weeks. In the meantime, the below is a good read.

go to post.

2015.05.11 biafra ahanonu 2015-05-11T00:00:00+02:00 2015-05-11T00:00:00+02:00

2015.05.11 [link]

Lighting the brain with optogenetics

Pretty interesting profile on Karl Deisseroth and the rise of optogenetics.

Rigor and reproducibility in scientific research

I recently posted a brief note on problems with statistics that can make a paper’s study hard to reproduce (see taking care with statistics), which is a widespread problem that the biomedical sciences are starting to grapple with better. If you haven’t seen it already, NIH’s site on reproducibility is worth a look. This is a nice addition and adds to the opinion pieces, such as Begley and Ellis’s highly cited warning, and efforts at magazine to have researchers put more information about the parameters and statistics used to design and analyze a study in print to help facilitate more replicable data going forward.

In addition Nature has compiled several articles and other resources into a single collection dealing with statistics for biologists.

update Was talking to several people. Pubmed Commons was suggested as a way to both improve reproducibility, provide an avenue for people to provide updates about a data set

go to post.

2015.05.08 biafra ahanonu 2015-05-11T00:00:00+02:00 2015-05-11T00:00:00+02:00

2015.05.08 [link]

Working with and sharing large datasets, part 1


Figure 7: A visual representation for one method of storing multi-dimensional, ever-changing biological data.[? ]

MathWorks has a nice summary page on dealing with large datasets in Matlab (Working with Big Data in MATLAB). This reminded me of an older article from the Sorger lab at Harvard that attempts to give a high-level view of how they manage data. It was that article along with several chats with people in my lab that convinced me to switch all imaging data from the TIFF file format to HDF going forward (has much better support for super large datasets), even if that made working with ImageJ more troublesome.

Remembering the reliance of people passing down or developing good data storage practices within the lab, I found a well-documented, standard resource outlining common practice for data storage in the sciences lacking. It would be nice to see more papers describing how people store their data and what techniques they use, e.g. CERN uses magnetic tapes for storage still vs. using hard drives. This, along with basic introductions to compression, would greatly save taxpayer and other money currently spent storing data either locally on hard drives and servers or off site on the cloud.

There are several papers out there describing efforts to standardize or share data in a particular field (e.g. fMRI data), but there doesn’t seem to be common guidelines for making this happen across neuroscience disciplines. Rather, each field (this applies to the biosciences in general) appears to have consortium or other efforts that come into vogue and either continue on through the efforts of several people or languish.

An argument could be made to create a government agency (or a group inside the NIH) whose sole purpose is to help manage and guide labs in storing data, deciding which bits (pun intended) to be shared, and the best way to describe the data (e.g. metadata) so future researchers can access and make sense of it. For example, NIDA (National Institute on Data Analysis)...but NIDA is taken by National Institute on Drug Abuse so better yet NIDSA (National Institute on Scientific Data Analysis). Some might complain that this is just adding another slow moving layer of bureaucracy, but it is sometimes shocking when you go to help other scientists analyze their data how non-standard formatting of names and other common variables is, especially by those who don’t program.

To this end, the NIH has several resources that have already implemented aspects of this vision. There is their main Data Science at NIH page, which has links to several other resources. In addition, several years back the NIH established the Working Group on Data and Informatics, which has been tasked (see NIH and Biomedical ‘Big Data’ along with DIWG's executive summary). The working group has several recommendations, some of which would be very useful if implemented broadly throughout the biomedical research community, such as a specific minimal set of metadata (and potentially specific templates for particular subfields). They also recommended to help develop quantitative training programs and identify gaps in terms of researchers trained in specific areas; though, I wonder how they will achieve the later goal.

In addition to the working group, there is also the NIH Big Data to Knowledge (BD2K) initiative, which launched in 2012 and seeks to better support digital infrastructure for managing, sharing, and utilizing big data. With the need for data scientists growing each year, efforts like this are welcome, but they appear to be largely focused on clinical or -omic (geonomic, proeteomic, etc.) datasets for the time being. Beyond the NIH, there are also other collaborative efforts moving forward, such as the Global Alliance for Genomics and Health.

In neuroscience in particular, there are projects already underway to start standardizing specific aspects of data collection and sharing, such as Neurodata Without Borders and others. There are several task forces at the International Neuroinformatics Coordinating Facility whose goal is to help develop standards for electrophysiology and neuroimaging data, but it is unclear how widespread adoption of their recommendations is or will be. Much like the NIH’s vision for data science, there will likely need to be a large push toward training young scientists early before they settle into bad habits or attempt to reinvent the wheel.

While projects like BrainFormat are great ideas and excellent to see, there is a particular problem inherent to academic research: many experiments are often single purpose/one-off affairs and the people involved might be more interested in getting a result and a paper than leaving behind an easily understood set of data for others to look through. This is a case where making data organization a part of the curriculum might accrue substantial gains down the road. This would at least ensure that people are aware of the resources available, past and current thinking, and areas for improvement. If this was a single day course, much like ethics is in many universities, this might be more easily adopted and would introduce a larger body of research scientists to tools needed to deal with the deluge of data, rather than each going off to their own lab to create ad hoc methods that suite their particular experiments.

Below are several links to articles, new stories, and resources related to formatting and standardizing big data sets.


The idea of modulating signaling pathways in a specific manner came up during lab meeting. Optogenetics has been the leading method for precise control of neural activity either by activating or inhibiting neurons via channelrhodopsin or archaerhodopsin. Another class of optogenetic tools exist known as optoXRs that splice a light responsive element to G-protein coupled receptors. This allows one to turn on intracellular signaling pathways, e.g. through cAMP, to modulate neural activity or other functions that might not just be represented in sub-second alterations in neural spiking behavior, as is done with traditional optogenetics. Below are a couple of papers that develop and utilize several different types of optoXRs.

go to post.

2015.04.30 biafra ahanonu 2015-05-11T00:00:00+02:00 2015-05-11T00:00:00+02:00

2015.04.30 [link]

3D CAD modeling of mouse surgeries


Figure 8: microCT of a rat head with blood vessels.

As the complexity of imaging experiments grows, it is becoming necessary to model out more and more of the entire experiment to ensure that you will have enough room on the mouse/rat’s head, that you can minimize damage and suffering to the animal, and provide the best data with the given setup. To do this, I’ve been using PTC Creo (though SOLIDWORKS and Autodesk's Inventor are other alternative CAD programs) to model mouse and rat surgeries. One problem is finding a suitable model of the mouse or rat brain to use. Luckily, several groups have conducted CT scans and other imaging to help render 3D models of different rodent brains. These can be imported, cleaned, and then used to design placement of different technologies to manipulate the brain, such as optogenetic fibers. See below links for a couple resources in this area.

go to post.

2015.04.22 biafra ahanonu 2015-05-11T00:00:00+02:00 2015-05-11T00:00:00+02:00

go to post.

2015.04.19 biafra ahanonu 2015-04-19T00:00:00+02:00 2015-04-19T00:00:00+02:00

go to post.

2015.04.18 biafra ahanonu 2015-04-18T00:00:00+02:00 2015-04-18T00:00:00+02:00

go to post.

2015.04.16 biafra ahanonu 2015-04-18T00:00:00+02:00 2015-04-18T00:00:00+02:00

go to post.

2015.04.02 biafra ahanonu 2015-04-04T00:00:00+02:00 2015-04-04T00:00:00+02:00

2015.04.02 [link]

Philosophical Transactions B review issue

Philosophical Transactions B has a good issue that has various researchers taking a look into the future of their respective fields, definitely worth a read.

Neuroscience Initiatives

The USA has the BRAIN Initiative, EU has the Human Brain Project, China has the Brainnetome initiative, and Australia has the AusBrain (though that is amorphous). As was noted a couple years back, `the brain is hot'. Philosophical Transactions B with overviews of the USA’s BRAIN Initiative and Japan’s Brain/MINDS project.

Defining consciousness


Figure 11: Overview of Integrated information theory’s ability to predict which systems can have experiences.

In Michio Kaku’s excellent book giving a high-level overview of neuroscience research, The Future of the Mind, he attempts to define consciousness in a very engineering-physics perspective: there are different levels of consciousness that are defined by the number of feedback loops that the system exhibits. For example, thermostats and plants are on the low end since they only have relatively simple feedback loops. Humans are a different class of consciousness because we are able to have multiple, parallel feedback loops going on simultaneously.

Giulio Tononi and Christof Koch attempt to define a theory, called Integrated information theory , that will allow us to explain consciousness. It is an interesting read, I’ve also included a link to a differing opinion as contrast.

go to post.

2015.03.28 biafra ahanonu 2015-04-04T00:00:00+02:00 2015-04-04T00:00:00+02:00

2015.03.28 [link]

trans-generational transfer of traits or Lamarck revisited


Figure 12: Description of transgenerational epigenetic inheritance.

I remember reading Dias, 2014. It starts to touch a bit on Lamarckism, or the idea that characteristics acquired by the organism. In school, we are classically taught that this hypothesis was incorrect and that mutations and natural selection are what determines which traits get passed on. Epigenetic Lamarckism is becoming more popular, as is the more neutral and precise term transgenerational epigenetics.

Below are a series of papers that cover epigenetic inheritance with some notes by me.



go to post.

2015.03.26 biafra ahanonu 2015-04-04T00:00:00+02:00 2015-04-04T00:00:00+02:00

go to post.

2015.03.20 biafra ahanonu 2015-04-04T00:00:00+02:00 2015-04-04T00:00:00+02:00

2015.03.20 [link]

Creating reward memories in the hippocampus


Figure 14: Protocol for inducing reward-related memories tied to a specific place by activating reward centers when place cells fire during sleep.[? ]

A paper recently showed that they are able to induce reward-seeking behavior in an animal by stimulating the medial forebrain bundle (MFB), which contain ventral tegmental area fibers that release dopamine, every time a place cell fires during sleep. This would imply that the place field still encodes the same information about an animals place in the environment even during sleep and hints that it is functionally meaningful. Would have been super cool if they were able to induce aversion as well.

One should compare this paper to previous false memory papers, though I like this one because it is manipulating behavior based on a single neurons activity, rather than broad spectrum reactivation.

go to post.

2015.03.15 biafra ahanonu 2015-04-04T00:00:00+02:00 2015-04-04T00:00:00+02:00

2015.03.15 [link]

Significance testing

While ago I wrote briefly about effect sizes and why they might be better than p-values, see statistics: effect sizes. It seems that Basic and Applied Social Psychology has decided to ban p-values (or null hypothesis testing) from the journal.

John Ioannidis has made a career from pointing out the troubling fact that there are way too many positive results in most scientific literature than should be expected by chance. This might partially be explained by the fact that people will continue to do a study until they obtain p

Miniature microscope and fiber photometry mini-review


Figure 16: Miniature microscopes and fiber photometry for measuring neural activity.

Garret Stuber has a minireview out in Nature Neuropsychopharmacology concerning miniature microscopes and fiber photometry. Similar to his paper from last year reviewing ways to image/manipulate neural activity.

go to post.

2015.03.12 biafra ahanonu 2015-04-04T00:00:00+02:00 2015-04-04T00:00:00+02:00

2015.03.12 [link]

SCAPE and other microscopy


Figure 17: Example of SCAPE microscopy, which both projects and images a plane of light in a sample through the same objective.[? ]

Elizabeth M. C. Hillman came by to give a talk on some of the work in her lab, which was quite interesting. She talked a bit about SCAPE microscopy, what influenced it’s development, and how they plan on improving it. She went over some of the fundamental limitations of point scanning two-photon microscopy along with other limitations of light sheet/field microscopy. The actual demonstration of SCAPE was fairly impressive, they were able to image in 3D whole Drosophila melanogaster larvae as they moved around a dish along with volumetric imaging in mice on a ball. However, when she showed a comparison to a two-photon volumetric image taken of the same slice, the SCAPE image appears considerably worse (both in the x-y, but it appears the z resolution degrades much faster with distance). But one assumes that this can be improved. Overall, pretty impressive.


Figure 18: Laminar optical tomography works by measuring the diffusion of light through a medium to nearby points. This can be used for volumetric reconstruction of images.[? ]

She also talked about laminar optical tomography, which appeared to allow her to do 3D volumetric imaging about a decade ago. Optical tomography takes advantage of the fact that light diffuses naturally through most medium, either tissue or artificial materials. Thus, if you shine light into a tissue, some of it will emerge at a distant location. The farther away it emerge, the longer distance it traveled and one can infer that this was likely in the z direction. Thus, by measuring the quantity of light emitted at points next to the spot of tissue being illuminated by the laser, one can estimate the density of material that the light traveled through and thus begin to reconstruct the 3D structure of the tissue being imaged. See below papers for detailed discussions.

She also showed a bit of data from her wide-field imaging setup to allow her to image both oxygenated and deoxygenated hemoglobin while also doing calcium imaging (using GCaMP). While it seemed most of the work was done in young animals, there was a strike lack of correlation between blood oxygen content and neural activity. If someone could at some point do a large scale fMRI and GCaMP (or use voltage fluorescent indicators) study of cortical activity, maybe that will finally cause people to start looking at fMRI studies with more skepticism, as it relates to claiming this or that brain region is involved in a behavior or cognitive process given it lights up on fMRI scans. Below paper covers this issue in a bit more depth.

Automatic detection of touch stimuli

This paper uses a nice method to automatically detect when touch stimuli have been applied:

There are various areas in the pain field, specifically stimulus delivery, that would benefit from automation that could allow more precise dissection of neural involvement in pain response/sensation when combined with new optogenetic and other techniques, seeoptogenetics and pain, continued and optogenetics and pain.

Puzzle imaging?

This seems to be along the lines of ‘expansion microscopy’ (Expansion microscopy, in that it is a novel name for a technique/algorithm that has already existed.

go to post.

2015.03.08 biafra ahanonu 2015-04-04T00:00:00+02:00 2015-04-04T00:00:00+02:00

2015.03.08 [link]

Fundamental limits of computation


Figure 19: As chips get smaller, they approach atomic limits, which leads to fundamental constraints on how densely packed CPUs can be with transistors while still being operational.

While back I posted about the limits of computation. Building off the discussion about AI (advances in artificial intelligence, part 2), there is an interesting question about whether superintelligence is fundamentally limited by thermodynamics and thus can only grow so far ahead of us. I’m not sure how optical or quantum computing invalidate conclusions of the below papers, but they are interesting reads nonetheless.

New mice from Allen Institute


Figure 20: More mice!

Allen Institute details a plethora of new mice in a recent Neuron paper.

However, I worry about GCaMP expressing mice because depending on how they are bred or what Cre line is used to drive expression, you could have an extra calcium binding protein in large numbers of cells during development and adulthood of the mice being studied. Whether this effects neural activity is unknown; though, the first place to look would be if there have been any calcineurin or other calcium binding protein overexpression studies in mice. It seems that people have performed these studies and seen changes in both cardiac function and memory. Some of the studies use modified forms of calcineurin (e.g. truncated) that may make extrapolation to constitutively expressed GCaMP problematic. However, it is worth giving pause, especially since no behavioral characterization of the mice seems to have been done in the paper.

go to post.

2015.03.03 biafra ahanonu 2015-04-04T00:00:00+02:00 2015-04-04T00:00:00+02:00

2015.03.03 [link]

3D imaging of zebrafish


Figure 21: 3D imaging of a zebrafish heart.[? ]

Fabian Voigt from Fritjof Helmchen’s lab came by last month. He demonstrated several videos from their new light sheet microscopes, which illuminate a plane of a sample to allow more rapid scanning of volumes. He showed several impressive videos of the beating heart of a zebrafish to demonstrate the technique.


Figure 22: Demonstration of how electrically tunable lenses work.

He also talked about electrically tunable lenses, which I ran into a couple times at Photonics West. These look super useful and allow rapid focusing without needing to move any mechanical parts, something that should both allow faster imaging of volumes while also reducing chance of system failure from wear and tear.

go to post.

2015.02.28 biafra ahanonu 2015-04-04T00:00:00+02:00 2015-04-04T00:00:00+02:00

2015.02.28 [link]

Advances in artificial intelligence, part 2


Figure 23: Example of an artificial neural network used to play Atari games from [? ]

Was talking with Benjamin Grewe, a postdoc in the Schnitzer lab, about the recent Nature paper from Google utilizing an algorithm developed at DeepMind. DeepMind was bought by Google and it seems this allows one to go from the arXiv to Nature. Bernhard Scholkopf gives a pretty good overview of the paper and some of the related literature. I’ll build a bit on my first AI post here, advances in artificial intelligence for the first entry.

This additionally got me thinking again about general artificial intelligence. Vernor Vinge and others have been predicting technological singularity or more accurately, the point at which better than human intelligence is created, for some time now. The issue still seems to be that most AI is domain specific. Even in the Atari example, that program would likely utterly fail if asked to interpret what is going on in the Oath of the Horatii or if it was asked to design a simple circuit to make lights blink when the door opens. How can narrow AI, that is AI which is very good at a specific task, be combined to make a more general AI that has order of magnitude more processing power and reasoning ability compared to a human working in the same field? Below are a number of articles exploring this issue in depth.


Figure 24: What happens when artificial intelligence doesn’t interpret incoming data in an intelligence, generalized manner. One of the largest intra-day drop in DOW history.

In addition, this in some sense alludes to another recent post on artificial intelligence and the law. What happens when an AI system commits a crime (either civil or criminal)? Should the programmer be punished or at some point would the AI be considered mature enough to itself be punished? How would you do this? Considering it could have backed itself up, shutting it down isn’t punishment enough. How would this punishment signal be sent to other AI systems? Would they care?

For example, Palantir, BlackRock, and others offer data analysis software. It wouldn’t be far-fetched to imagine Google allowing one of DeepMind’s algorithms to take these company’s data analytics frameworks and enhance them for use in the stock markets. It could look at historical stock data, gather information on companies HQ and other infrastructure location, and correlate all of this with up-to-the-minute news via Google News. Thus, because they would have information on the types of words being used previously when a stock crashed or surged (via Google Trends), e.g. if there was an increased correlation of a company’s name in the news with negative wording, they could potentially take advantage of large fluctuations better than current high-frequency algorithmic trading. This would obviously only work for a short period of time, similar to the initial fabulous gains by some participants in the 80s and 90s when computers/rigorous statistics were first coming into play. But it could also lead to situations like the Some articles exploring that event in more detail below.

go to post.

2015.02.19 biafra ahanonu 2015-04-04T00:00:00+02:00 2015-04-04T00:00:00+02:00

2015.02.19 [link]

On publishing about what we already knew


Figure 25: Left: optogenetics paper from 2015[? ]. Right: rat stimulation paper from 1982[? ]. Both demonstrate that the SCN can influence circadian cycles by stimulating it (blue shared region in left, arrows in right image).

Manipulating circadian clock neuron firing rate resets molecular circadian rhythms and behavior -Really, the SCN is important for ? I immediately was slightly suspicious since this seemed to be common knowledge and something learned in Neuro201. Except when we learned about it, they used electrical stimulation. After a single PubMed search, I came across one of the older 1980s papers that basically already proves what this paper is demonstrating. Additional review article from the 1970s included for additional reading.

To this end, DrugMonkey has a rather biting post on a recent Nature paper, see Nature publishes overwhelmingly proven ``NEW AMAZING FINDING''....because optogenetics!. Anyways below is the new Nature paper and a cadre of older papers looking at the behavior investigated.

Also, I noticed that Charles Zuker has seemingly only published in Cell/Nature/Science over the last decade (see Zuker lab publications). Didn’t know that was possible.

Imaging axon activity with SyGCaMP

Imaging activity in axons will be a crucial step in understanding how information between brain regions is transfered. While somatic activity is important, that signal is ultimately conveyed via an axon and the micro-environment around the axon terminals in different brain regions could potentially lead to the same somatic activity having differential pass on that information to various downstream brain regions.

It is currently extremely difficult to measure electrical activity in an axon coming from a specific brain region. For these reasons, fluorescent protein indicators of neural activity are used, such as GCaMP. One can do this be either expressing GCaMP at high levels such that it diffuses down the axon or by tagging it with synaptophysin GCaMP (SyGCaMP) to help localize it to axon terminals. This later strategy has been used before and seems promising for future work, see below for a couple articles.

go to post.

2015.02.14 biafra ahanonu 2015-02-15T00:00:00+01:00 2015-02-15T00:00:00+01:00

2015.02.14 [link]

The trend towards funding older investigators


Figure 26: The average age of NIH investigators under 36 years old continues to decline.

There is a rather worrying trend toward National Institute of Health (NIH) R01 grants (one of the main funding mechanisms used by the NIH to fund research) going to older and older investigators. While this helps maintain large labs and fund a variety of important projects, there is an argument to be made that this money is not being spent in the most effective manner.

This argument can come from several angles, the most obvious is using the criteria of scientific productivity. Using Nobel Prizes as the gold standard, across biology, chemistry and physics the most productive years are between 34-38 years.[? ] Using other criteria, it was found that productivity of biomedical scientists appears to peak in their early 30s.[? ] However, when one looks at the age distribution of NIH R01 investigators (Fig. ??), the number of faculty and new investigators getting grants in this age range is dropping fast. However, it is unlikely that this will change unless there is a revolution in the NIH’s incentives and structure. I cannot find data listing the ages of the NIH review committees, which might provide some insight into the biases that might be driving this trend.

Below are a list of several publications that look at scientific productivity as a function of age:

This also gets into questions how the current state of postdoctoral research, but I will save that for another time. In the meantime, below are a couple interesting reads.

Imaging neural activity in 3D


Figure 27: Several ways to do volumetric imaging.

Last year I discussed several papers that imaged zebrafish using light sheet microscopy, see whole animal 3D imaging. Now that Misha Ahrens has his own lab at Janelia, he has advanced the work he published two years ago. He has an excellent primer in Neuron going over different methods to visualize animal neural activity in 3D.

Determining anatomical locations or how well does animal work replicate human physiology and brain activity?

9.14 - Brain Structure and Its Origins, taught by Prof. Schneider at MIT, was a great comparative anatomy course looking at how the brain evolved over time and the similarity and differences between various species’ brains. One topic that came up is what areas could be considered equivalent between primates, other mammals, and birds. To this end, the below paper is an interesting read to get an idea how one might argue that a particular brain region known to exist in humans also exists in lower mammals, in this case the prefrontal cortex in rodents.

This also gets at another issue, which is how closely findings in lower animals, such as the widely used rodent, match similar studies in humans. There are many arguments both in favor and against animal models, from the number of pre-clinical animal model drug studies that don’t end up having the same effect in patients. There need to be more studies like Becerra, 2013 that measure experimental variables in both humans and rodents to help assuage fears that rodents neural activity is fundamentally different in response to pain.

go to post.

2015.02.12 biafra ahanonu 2015-02-15T00:00:00+01:00 2015-02-15T00:00:00+01:00

2015.02.12 [link]

GPS tracking in bats


Figure 28: Example of GPS tracking in bats.[? ]

I previously talked about spatial navigation in bats and while at a graduate student seminar by Hannah Frank in the Hadly Lab, the topic of GPS navigation in bats came up. I vaguely remembered that the Ulanovsky Lab in Israel had done similar work and it seems that is actually the case. See below for a couple articles on the subject. It would be amazing if someone could record from the bat’s hippocampus during these long flights to see how spatial information is represented on these scales, e.g. do place cells expand the size of their place fields, respond to many more places along the bats path, or does a higher-level code relating the activity of multiple place fields kick in?

Improving PCA-ICA for cell detection in calcium imaging

There are a number of techniques to detect cells using calcium imaging, see calcium imaging cell detection techniques. PCA-ICA is among the algorithms used, but can be slow and problematic on larger matrices. The below paper proposes a GPU based method to speed it up, which could enhance methods that use this for cell detection.[? ]

Human Brain Project

If you haven’t already, take a look at the EU’s Human Brain Project’s webpage. The merits of the program will be saved for another time.

go to post.

2015.02.11 biafra ahanonu 2015-02-15T00:00:00+01:00 2015-02-15T00:00:00+01:00

2015.02.11 [link]

Virtual reality and BMI


Figure 29: Concept photo for Microsoft’s HoloLens.

In some sense, one end goal of neuroscience would be to understand the brain enough—both how it is wired and the pattern of neural activity used to produce sensations, store experiences, and create ideas—to allow us to manipulate its activity to allow for entirely new experiences. And while virtual reality is currently very hot in tech—Facebook bought Oculus, Google Glass is still chugging along (and in need of a revamp or new marketing), and Microsoft recently unveiled HoloLens—a real advance would be the ability to directly project scenes and experiences into the human brain, such as a recent study that used electroencephalography (EEG) to record brain signals from one human and use them to induce motor behavior in another human via transcranial magnetic stimulation (TMS) of the motor cortex.[? ]

On a related note, this would potentially allow us to hijack our knowledge of the brains amazing plasticity to expand our repertoire of senses. Neil Harbisson gave a talk about listening to color, whereby a device allows him to hear different colors, see I listen to color. This uses an eyeborg that converts light into sound. It is easy to imagine adapting such a device to list in for UV, X-ray, Infrared, and other radiation, as is used in a primitive manner with Geiger counters and similar devices. Only, imagine this information could be routed directly to the visual cortex, allowing one to rudimentary see other forms of radiation without blocking visualization of the normal visible light spectrum.

On writing well

Foreign Policy has a great article about academic writing. At the end he mentions George Orwell’s great piece Politics and the English Language, which I’ve always found to be a great primer on the reasons why writing clearly is crucial for maintaining logical clarity. Below are some additional pieces on science writing.

The article recommends a couple books for those looking to write clean, concise prose that a reader can easily and logically follow.

He also notes several scientific authors whose writing abilities are to be admired and studied: Waltz, Thomas Schelling, James Scott, John Mueller, Deirdre McCloskey, and Charles Krauthammer.

Photonics west

Went by Photonics West 2015 yesterday with Tony Kim, another graduate student in the Schnitzer lab. If you are ever in the Bay Area when it is on, you should at least go by the exhibition. There are a plethora of companies who come each year and you are bound to find something new and interesting, in addition to getting the opportunity to talk with people from specific companies if you have questions.

go to post.

2015.02.01 biafra ahanonu 2015-02-15T00:00:00+01:00 2015-02-15T00:00:00+01:00

2015.02.01 [link]

Investigating hypothalamus activity with the miniature microscope


Figure 30:

Garret Stuber has a new paper detailing the role of specific lateral hypothalamic (LH) neuron populations in promoting appetitive behavior. Part of the study uses the miniature microscope to image endogenous neural activity in the LH. The resulting analysis is rather simple, but does suggest that consummatory and appetitive behavior induce firing in non-overlapping LH neuron populations.

Their videos have a good deal of brain motion, which will affect proper cell detection and calcium signal extraction techniques; though, their neurons are magnified (owing to the fact that they are using the smaller 0.5mm in diameter lenses instead of the more standard 1.0mm). This is an okay demonstration of the technique, no detailed analysis of the code used by the LH beyond some cells go up and others go down or validation of their cross-session cell alignment is given. Also, the classification of cells into Food-zone excited and Food-zone inhibited cells is not very precise, they only use a ratio between calcium activity in the two zones rather than mutual information or a more precise metric (that takes into account firing rates, time spent in location/how often stimuli are presented, etc.).

A quick note: Figure S6 is rather misleading, as they don’t plot it compared to the animals activity and thus it is unclear whether cells become less responsive to the task or the animal just stops moving (they don’t have a figure plotting a cumulative total movement or distance per time bin as is standard when tracking animal movement in an open field-like environment).

Anyways, as more papers are published using this technique, the rigor in analysis will likely go up as the wow factor decreases and people start looking more at the content of data being produced by the technique.

As a nice bonus, the results match what was seen in a paper in the same issue from Kay Tye’s lab.

go to post.

2015.01.30 biafra ahanonu 2015-01-30T00:00:00+01:00 2015-01-30T00:00:00+01:00

2015.01.30 [link]

I will likely turn the below post into a permenant resource here on my main website.

Advice for starting graduate students: choosing a lab

Choosing a thesis lab, advisor, and scientific problem you want to solve can often be a confusing experience without some guidance and considering that some decisions will end up defining one aspect of your life for several years, it is important to spend the time to get a variety of advice from as many sources as possible. Below are articles going over this decision and what to look out for during rotations and other interactions leading up to deciding on a lab.

Advice for starting graduate students: being successful

Sometimes it’s good to go back and read the documents you browsed through when first starting graduate school. The main reason is that helps provide perspective and identify where you might be going astray. Below is a useful list of resources along the lines of what it takes to be successful in graduate school.

go to post.

2015.01.29 biafra ahanonu 2015-01-30T00:00:00+01:00 2015-01-30T00:00:00+01:00

2015.01.29 [link]

Does science require experiments?

That is one debate going on in physics as people start to take string theory, multi-verses, and other hypotheses more seriously. The key problem is that some of these cosmological hypotheses are not testable, at least not given what we know about the current rules of physics. This raises a key question, if a theory is sufficiently elegant and mathematically sound, should one need to conduct experiments to prove it? As a biologist, I’m heavily biased towards yes, as that is the foundation of science. However, there are arguments in the other direction as well. See below.

Photoactivatable neuropeptides

While the ability to image neuropeptides directly would be super useful to understand how changes in their local concentrations within a brain area contribute to differential activity dynamics of neurons in said brain region, photoactivatable also have their uses. See below paper.

go to post.

2015.01.28 biafra ahanonu 2015-01-30T00:00:00+01:00 2015-01-30T00:00:00+01:00

2015.01.28 [link]


Pretty good overview of Sebastian Seung’s work on connectomics. Can also watch his TED talk for another layman’s overview of the field and its broader goals.

Haruhiko Bito talk

Haruhiko Bito gave a talk today that covered some of his lab’s work using an improved version of RCaMP, a red fluorescent indicator of calcium concentration that is used as a readout of neuronal activity, to do simultaneousness imaging in cortical somatostatin interneurons and excitatory pyramidal cells (using both RCaMP and GCaMP). While the same could be accomplished genetically by crossing Cre versions of those mouse lines with tdTomato reporter lines, this might open up additional experiments that wouldn’t be possible, given that Cre expression during development isn’t always specific, etc.

Science magazine issue on ESA’s Rosetta mission

It is still mind boggling the amount of technical skill needed to pull off the Rosetta missions, e.g. the precision involved in allowing two small objects to rendezvous in the vacuum of space.

go to post.

2015.01.22 biafra ahanonu 2015-01-30T00:00:00+01:00 2015-01-30T00:00:00+01:00

2015.01.22 [link]

Gyorgy Buzsaki talk

Gyorgy Buzsaki gave a talk at Stanford today that was fairly enlightening. I’ll just link to several papers he mentioned that are good reads.

dual photostimulation and optogenetics

Might as well post the full list of relevant articles relating to this sub-field that is bound to explode (I’ve already talked about it: see dual photo-stimulation and imaging, freely moving dual photostimulation and imaging, and dual photo-stimulation and imaging, cont'd)

go to post.

2015.01.20 biafra ahanonu 2015-01-30T00:00:00+01:00 2015-01-30T00:00:00+01:00

2015.01.20 [link]

10,000 bit word computers

This is an interesting paper that attempts to look at how a 10,000 bit computer would be designed. Keep in mind most computers today are either 32-bit or 64-bit (and in many applications that seems more than enough).

Theoretical neuroscience issue

Came across this issue of Current Opinion in Neurobiology awhile back, but still find it quite useful.


Groups such as Jin Hyung Lee at Stanford and others have looked into using optogenetics and fMRI to probe brain wide neural response to specific perturbations. This could lead to interesting analysis of the effects of stimulating neuromodulatory brain areas—ventral tegmental area, locus coreleus, dorsal raphae, etc.—on whole brain activity. However, always keep in mind that fMRI currently does not measure neural activity, but blood oxygen content, which has been shown in various cases to be weakly correlated, or not at all, to neural activity.

Ecological expected utility and the mythical neural code

Interesting read, looks at re-framing the issue of what the neural code is, e.g. how we make sense of the firing patterns in the brain as they relate to behavior or other cognitive outputs/processes.

go to post.

2015.01.18 biafra ahanonu 2015-01-30T00:00:00+01:00 2015-01-30T00:00:00+01:00

2015.01.18 [link]

Human optogenetic trials

It seems we are seeing the first optogenetic human trials coming into play. They are going for one of the more obvious applications and hopefully one that provides successful. The plan is to use channelrhodopsin expressed in blind patients retinas to help them gain crude visual capabilities. See below for more detailed articles on the subject.

Also the rest of the NewScientist 2015 science stories are interesting:

Another look at the Turing Test

An interesting paper looking at whether machines can think and how the Turing Test has been misunderstood. This paper discusses more of the nuances of Turing’s thought experiment that helps clarify confusion about what the test does and does not tell us about intelligence of another collection of negative entropy matter (i.e. life, see Schrdinger’s What Is Life?).

Something to keep in mind now that AI is growing exponentially (see DeepMind) and that if the EU succeed in its initiative to simulate the human brain, the question will arise how do we test that it is ‘intelligent’ in our definition of the term.

go to post.

2015.01.16 biafra ahanonu 2015-01-30T00:00:00+01:00 2015-01-30T00:00:00+01:00

2015.01.16 [link]

automating medical health records

Troy Astorino is a friend of mine from MIT who started PicnicHealth (see PicnicHealth Stores Your Medical Records In One Place And Delivers It To Your Doctor). He mentioned that the electronic medical health market is still far from being standardized in a proper way. For example, Health Level 7 is a supposed standard for transferring medical records between hospitals, but each hospital uses its own custom version of the standard, it apparently isn’t the best standard, and can be costly to implement. Anyways, this market will continually heat up and it will be interesting to see whether and how standards like those seen in the tech hardware and software industries emerge.

random person to know: David Hilbert

David Hilbert. That is all.

linguistic relativity

Linguistic relativity is interesting conceptually and I’ve often wondered how much thought is actually constrained by the language we know. In some sense, it is often hard to think abstractly without resorting to internal monologues in your language of choice. This contrast to abstract thinking in pictures or symbols, which should be much less constrained by your chosen language. It also begs the question of whether as a language becomes corrupted, e.g. as words begin to lose their specific meaning as George Orwell commented on in Politics and the English Language, the actual thought process and logic of individuals using that language does as well.

An interesting related question is whether our brains store verbs, nouns, adverbs, and other grammatical constructs differently within the brain, e.g. is the firing patterns in the auditory cortex or Wernicke’s/Broca’s in response to and during the production of speech more distinct between nouns/verbs or solely based on the frequency spectrum produced by those words? Would maybe help answer questions posed long ago about why some word categories are learned sooner than others, e.g. see Why nouns are learned before verbs: Linguistic relativity versus natural partitioning.

ultra fast imaging


Figure 31: Setup for imaging 1011 frames per second.

From a purely technical standpoint this is an interesting article. That they were able to visualize single laser pulses reflecting off of mirrors and being refracted is super cool. Would be informative to combine this with recent advances in metamaterials that have allowed invisibility to become a reality on the macroscopic scale to look at how light travels through those materials, if possible.

go to post.

2015.01.14 biafra ahanonu 2015-01-30T00:00:00+01:00 2015-01-30T00:00:00+01:00

2015.01.14 [link]

high throughput screening of C. elegans


Figure 32: High throughput screening of C. elegans.

At the CNC Annual Symposium last year Hang Lu gave a talk about a collaboration with Kang Shen to help develop high throughput microfluidic devices for screening synaptic deficits in C. elegans. One key take away from the presentation was that machines could detect more subtle changes in synapse morphology, position, and other changes than humans. Further, they could detect changes that induce many weak changes among a series of metrics, something humans find extremely difficult to assess in a rigorous, quantitative manner.

Continuing this theme, at the 2013 Conte Center Neuroscience Symposium, Ulrike Heberlein talked about work she was doing on quantitative descriptions of Drosophila courtship and other behaviors using machine vision in collaboration with the Branson Lab. Along similar lines, another lab at Janelia recently published a high-throughput method of screening for neurons involved in particular Drosophila behaviors (see Discovery of Brainwide Neural-Behavioral Maps via Multiscale Unsupervised Structure Learning). Whether these types of high-throughput characterizations of animal behavior and circuits yield novel or specific insights (beyond just increasing the complexity of the problem) still remains to be seen. Excited to see similar methods applied to research in rodents, which is oftentimes too focused on a single species (Mus musculus or Rattus norvegicus) and a narrow, specific behavior. Curiously, this same complaint about biomedical research has come up more than half a century ago, see Beach’s classic .The snark was a boojum[? ]

go to post.

2015.01.12 biafra ahanonu 2015-01-30T00:00:00+01:00 2015-01-30T00:00:00+01:00

2015.01.12 [link]

the stress of biomedical research

There are many stressors in biology, from experiments that don’t work to charting out the next steps in one’s career. While many of the stressors are common to any job, several are unique to biology because the system being manipulated and studied isn’t the creation of man. And this causes it’s own peculiar problems for our control-obsessed human minds. The below article goes over some common issues biologists encounter while conducting research.

3D place fields in bats


Figure 33: Comparison of spherical vs. toroidal systems for encoding bat’s position and orientation in space.

I remember several years back seeing a presentation on bat 3D place fields at Janelia Farm. You know a presentation is good, or the finding novel enough, that you can remember the room you were in while watching it over three years later. This work comes from Nachum Ulanovsky’s lab, who specializes in studying the coding of space in bats. In the most recent paper they demonstrate that head direction coding of the bat is consistent with a toroidal model of the world, that is if you model the bat’s yaw axis as rotating around the major radius and the pitch as rotating about the minor radius, then this accounts for the patter of place cell activity seen (see figure above or Extended Figure 7a in the text). It also means that the orientation and direction of the bat matters, which would not be the case in a spherical configuration.

random cancer

A new paper claims that lifetime risk for getting a particular cancer is highly correlated with the amount of stem cell divisions that a specific tissue undergoes. On the face of it, this makes sense because each time a division occurs, there are chances for mutations to take place that lead to cancer. This has important health implications, as it can imply where treatment should focus, e.g. some cancers might inherently be more treatable through traditional means. This reminds me of an older paper by Piyush Gupta, my old academic advisor at MIT, showing that cancer populations will settle on an equilibrium state. This had implications for treatment, because it suggested that no matter how much you tried to eradicate a particular cancer, it would always come back given the right environment.

go to post.

2015.01.10 biafra ahanonu 2015-01-30T00:00:00+01:00 2015-01-30T00:00:00+01:00

2015.01.10 [link]

physicist view of the world


Figure 34: Comparison on a biologist’s (A) and engineer’s (B) view of how a radio works, from a physicist perspective.

Was recently listening to Elon Musk talk about his work ethic and principles (see Elon Musk - Work ethics, Principles, Attitude, Failure - Pearls of Advice). In it, he reiterated a point that one notices him mention in many of his interviews and speeches, the idea that taking a first-principles view of the world, as physicists are taught to do, rather than working by analogy can allow you to discover new truths by pointing out logical gaps or counterintuitive thoughts that others missed. He often gives this advice when talking about the formation of SpaceX, that he first reduced the problem of making a space craft to it’s simplest form, that is the raw materials. He then worked up from there, estimating the costs of each component and realizing that rockets in essence where cheap, it is just the rearrangement of the raw materials that is expensive (e.g. like much of manufacturing).

This lead me to start searching for other cases where people attempt to generalize from a physicist’s point of view and apply it to the real world. Below are a collection of articles along those lines.

viral tropism

We had a discussion at a lab meeting about viral tropism. In particular, there was the worry that specific retrograde viruses (e.g. rabies, psuedo-typed rabies, CAV, etc.) may display cell-type specific expression patterns and propensity to travel. Below is a paper that describes the transduction efficiency, expression patterns, and other aspects of various adeno-associated virus (AAV) serotypes. A useful reference.

go to post.

2015.01.08 biafra ahanonu 2015-01-30T00:00:00+01:00 2015-01-30T00:00:00+01:00

2015.01.08 [link]

freely moving two-photon imaging


Figure 35: One configuration for two-photon freely moving imaging.

Was looking into two photon imaging in freely moving animals, it would be useful for multi-colored imaging of different cell-types or other experiments. However, rather than finding a glut of papers, it seems that this fell out of popularity. Several lab members noted that this was partially owing to the fact that instead of using galvos to scan the field of view, most designs vibrated the fiber, which leads to many problems when an animal is moving and is apparently not very reliable. Further, the field of view was never very large, limiting its usefulness in studying large neuronal populations. Anyways, a couple of the papers below:

Bongwoori, a new voltage sensor

I recently pointed to several papers concerning voltage imaging (see voltage imaging, screening for better probes). A new probe can be added to the list, it is a modified version of ArcLight called Bongwoori. The paper talks a bit about the process they went through to create the probe and go on to characterize it’s properties in hippocampal slices. Seems promising.

imaging multiple brain regions

On a slightly different note (building off the voltage sensor news above), there has been a trend toward mesoscopic imaging (see large scale imaging and other papers) and this paper extends that to the visual cortex using voltage indicators. It would be particularly interesting to see someone combine two colors of voltage imaging to look at how membrane fluctuations of glia/astrocytes influence neuronal membrane potentials on a global scale.

A paper released last November by Jrme Lecoq, a postdoc in my thesis lab, uses multiple microscopes to image distal brain regions. While not being able to cover as wide a field of view, he was able to do two-photon imaging and this would allow imaging of interacting brain regions that might be hard to cover in a single field of view.



Figure 36: A nice overview of molecular and cell biology related institutions in China and when they were founded.

China is one country where I am still unsure about the status of the research there, e.g. how much of it should be trusted and what the general direction of research there is. For example, while Singapore started off with a fervent passion for basic biomedical research with relatively few top-down restrictions (see Singapore's salad days are over), that began to change. I remember Ian Cheong and others at Temasek Life Sciences while I was there during the summer of 2012 mentioning that there was a shift toward more applied research. Whether that is happening in China and how the vast funds at the government’s disposable are being put to use is still unclear to me. Hopefully I can clear that up soon. In the meantime, this article provide a nice overview of the state and history of molecular and cell biology research. There should be more articles like this on other countries. Especially concerning the pharmaceutical industries in various countries (this probably exists, just need to find an appropriate report).


For those wanting to do any work with microcontrollers, the Intel Galileo seems like a good option.

go to post.

2015.01.06 biafra ahanonu 2015-01-06T00:00:00+01:00 2015-01-06T00:00:00+01:00

2015.01.06 [link]


Figure 37: A setup to allow optogenetic identification of single units during tetrode recordings.

optogenetics identification of single unit recordings

While there are many advantages to imaging—being able to track cells across days, genetic specificity, dual photostimulation and images, and many others—the temporal resolution and several other advantages still make tetrodes a useful tool in any systems neuroscientist’s belt. Below are a couple papers that utilize a technique whereby channelrhodopsin (or another optogentically activatable protein) is expressed in a subset of neurons of interests. Shining light while recording neural activity allows one to identify units that are light responsive, and are thus part of the genetically defined subset that channelrhodopsin was expressed in. Depending on the location and type of cell, there can be asterisks associated with this technique, but it seems to be picking up steam as a viable alternative to calcium imaging, especially as voltage imaging isn’t quite ready for prime time in vivo. Below are some relevant articles in the area.

go to post.

2015.01.05 biafra ahanonu 2015-01-05T00:00:00+01:00 2015-01-05T00:00:00+01:00

2015.01.05 [link]

optogenetics and pain, continued

While back I linked to several articles that talked about optogenetics and pain (see optogenetics and pain). Wanted to link to several more papers, especially as this particular sub-field has progressed quite a bit since then. In addition, I’ve included several short overviews of the field and a paper that looks at using the technology to restore muscle function after peripheral injury. I’ll talk at length about this area more in the future, but it could help advance the pain field significantly in the coming years while providing hints at possible therapeutic avenues for chronic pain in the coming decades.

go to post.

2015.01.04 biafra ahanonu 2015-01-05T00:00:00+01:00 2015-01-05T00:00:00+01:00

2015.01.04 [link]

advances in artificial intelligence

Was recently watching Deepmind artificial intelligence @ FDOT14, where they demonstrate a new machine learning algorithm that can learn arbitrary rules that govern the movements of pixels on a video screen, also known as learning to play a video game. What is amazing about this is that they don’t give the machine syntactic rules about the games, only the raw video feed and access to a controller, much like a human would have. After an hour or more of training the machine is performing at or above human levels in a variety of Atari games (Pong, Space Invaders, etc.). The white paper can be found at Playing Atari with Deep Reinforcement Learning (also points out the asterisks associated with the algorithm, but still cool nonetheless).

This was partially spurred by recent comments Elon Musk has made regarding the potentially exponential growth in artificial intelligence that could pose a grave danger to humanity in the future should it go unchecked. Nick Bostrom has a rather enlightening book called Superintelligence Paths, Dangers, Strategies that gives an overview of the field of people looking to understand how superintelligence among artificial intelligence might emerge and what we can do about it.

These developments are interesting because it also shines light on a fundamental goal of neuroscience: discover what algorithm the brain uses to allow processing of sensory input to produce an optimal behavioral output. Presumably this would allow us to build better robotic devices, improved search and other algorithms, and develop a host of new technologies. To put things in perspective, a small fruit fly no smaller than a lowercase ‘o’ on the keyboard can navigate complex environments and avoid potential predators using minuscule amount of energy. We currently have nothing machine-wise that can perform at that level and ones that can need orders of magnitude more energy. However, if more and more generalizable algorithms like DeepMind come into play, it may obviate the need to even explore this aspect of neuroscience. This is something people studying reinforcement and other learning paradigms should keep in mind when selling their research, though the issue is still probably a decade or two away from becoming serious enough to cause people to start looking for other jobs.

artificial intelligence and the law

On a drive down from northern California awhile back Devon Chandler-Brown and I had a conversation about what would happen if a robot committed a crime, say murder or burglary? Who would be responsible? If the owner was responsible, that would open up a whole can of worms, because in essence that would be punishing the parent for the sins of their offspring, something we’ve in general moved past (in the West). However, if the robot was to be punished, how would that play out? If they are not fully self-aware, shutting it down doesn’t send a negative signal to other semi-aware robots (as execution or imprisonment does to humans). On the other hand, if they are fully-aware, they would probably have backed themselves up somewhere, so any punishment is moot since they could just boot up the back-up in case of emergency.

Rather than going into the minutiae of that conversation, I would like to point to several articles that go over this issue in greater detail, as it seems that lawyers are now becoming aware of this fascinating topic and trying to see how it fits within current law.

curing Alzheimer’s?


Figure 38: EP2 pathway can have a number of downstream effects.

The popular press is at it again, making bold claims about EP2 research scientists at Stanford have performed. While neurodegeneration is undoubtedly a pressing problem, a healthy skepticism should be made about cures after a recent spat of clinical failures, see the γ-secretase Alzheimer's trials and others. The problem with the γ-secretase trials was that you had to make sure Notch signaling wasn’t radically altered, seeing as it is important signaling cascade. The same might be true for EP2 (see above figure).

go to post.

2014.12.24 biafra ahanonu 2015-01-05T00:00:00+01:00 2015-01-05T00:00:00+01:00

2014.12.24 [link]

astrocyte imaging

I’ll do a larger post on this later, as imaging glia and correlating their population responses with neural activity seems to be a wide open frontier in neuroscience, but for now the following paper is an interesting read.

voltage imaging, screening for better probes

Another topic that is ripe for review, earlier this year Yiyang Gong, a former postdoc in the Schnitzer lab who now has his own lab at Duke (check out his lab's website), published an improved voltage sensor called MacQ that improved upon previous voltage sensors (such as Arclight). In addition, the Cohen and Boyden labs have developed QuasAr, an improved voltage sensory, and CheRiff, an channelrhodopsin variant for use with QuasAr in dual stimulation/imaging experiments. Screening for new voltage probes, especially ones with higher signal-to-noise that require less laser power, would be boosted by improved high-throughput methods for doing so. The below paper does some initial characterization in this area.

single cell analysis

Along the lines of improving voltage sensors, it seemed that taking a page from single-cell analysis field might provide opportunities to improve throughput while increasing the sensitivity of the procedure. I would imaging a microfluidic device that one inputs different clonal populations (of mutations from base voltage sensory). At the middle is a chamber where a single cell is held and a small, consistent voltage applied. A microscope records the resulting change in fluorescence and decides whether this cell meets criteria, if so it is sorted in one direction and plated into a single well in a 96-well plate. This would be repeated and after those clones grew and a new round of mutants made, the process repeated. Anyways, below are a couple papers that give an overview of the field or are primary articles.

go to post.

2014.12.22 biafra ahanonu 2015-01-05T00:00:00+01:00 2015-01-05T00:00:00+01:00

2014.12.22 [link]

dual photo-stimulation and imaging, cont’d


Figure 39: Setup for dual photostimulation and imaging in Packer, et al. 2014.[? ]

Last month/week I talked about papers from the Tank lab and Emiliani group detailing dual photostimulation and imaging of neural activity (see dual photo-stimulation and imaging and freely moving dual photostimulation and imaging). Adam Packer, from Michael Hausser’s lab, has published a paper in Nature Methods adding to this growing aspect of neuroscience research.

Also, the below papers might be of interest as further reading. They demonstrate earlier uses of both patterned excitation profiles with SLMs and laser-scanning excitation.

dual photo-stimulation and imaging, update


Figure 40: left, from Adam’s paper (supplementary figure 1) measuring the probability of a spike on off-target neurons as a function of distance from photostimulation spot. middle, similar as left (supplementary figure 5), only with increase numbers of photostimulation spots. right, from Tank paper[? ] (figure 3c) showing evoked responses as a function of distance from illumination spot.

Asked Adam about the issues of non-specific photostimulation that both him and the Tank lab are seeing in these experiments, I’ve summarized his response below.

The most relevant figures are figure 3b in the Tank paper and supplementary figures 1 and 5 in Adam’s (see above figure). I had originally thought that the spatial light modulators (SLM) that Adam and co. used in their paper might increase the PSF size leading to reduced specificity in the targeted illumination pattern. However, he mentioned that was not the case and that their PSF size is still determined by objective filling parameters (see The Diffraction Barrier in Optical Microscopy for more). He indicated that it is still unclear how much non-specific stimulation is due direct photostimulation or evoked responses from neurons connected to the targeted neuron.


Figure 41: left, stimulation by dendritic processes.[? ] right, out-of-focus channelrhodopsin excitation.[? ]

He did note that photostimulation of dendritic processes was possible (see figure 2 of Packer, 2012) or that out-of-focus stimulation could occur due to channelrhodopsin’s large two-photo excitation profile (see figure 2 of Rickgauer, 2009).

I’ll provide additional updates/thoughts that seem relevant/pertinent as they arise.

go to post.

2014.12.18 biafra ahanonu 2015-01-05T00:00:00+01:00 2015-01-05T00:00:00+01:00

2014.12.18 [link]

neuroscience foundations and institutes

The BRAIN initiative provided added fuel to an already growing fire, in this case the trend in neuroscience toward larger collaborations as it has become clear that to tackle the complexity of the brain, expertise from a larger variety of fields needs to be coordinated toward a common purpose. While the benefit of Big Science may be in laying out groundwork for small science to make conceptual advances (as Graur and co. argued was what went wrong with the ENCODE project [along with others], since it tried to do both), there is still the possibility that mixing the two in a single institute could yield a higher probability that the transition would take place. Below are a couple of the new foundations/institutes being pushed along with a slightly older one.

using stem cells to treat eye degeneration

Several years back the FDA created the Breakthrough Therapies program to help speed the approval process of drugs that will demonstrably provide dramatically better care for a specific group of patients. While the below trial is only a demonstration of using stem cells to treat age-related macular degeneration, it would be interesting to see whether the results from these types of early-stage clinical trials can be fast-tracked into hospitals, should they prove to be safe.

CaMPARI update


Figure 42: Cells expressing CaMPARI show different ratios of red/green depending on their activity at the time UV light was shone on the zebrafish.

Earlier this year I reported on a talk by Loren Looger that detailed some new technologies coming out of Janelia (see new neuroscience tools from janelia). Below is a short piece talking in more detail about CaMPARI, the .

go to post.

2014.12.15 biafra ahanonu 2015-01-05T00:00:00+01:00 2015-01-05T00:00:00+01:00

2014.12.15 [link]


Figure 43: Outline of research experiment using the miniscope to test effects of Zolpidem on neural activity.

new miniature microscope papers

Now that the miniature microscope is in the hands of many labs we should start seeing many of its potential promises: imaging of neural activity in freely moving mice, tracking of neurons across sessions (e.g. days), exploration of genetically or anatomically defined subsets of neurons, and more. A couple of miniature microscope papers have been released recently. One details the effects of an GABA-A agonist (Zolpidem, normally used to treat insomnia) on CA1 neural activity. It would have been more interesting if they also imaged GABAergic neurons in the CA1 to see how they are affected as well. The other paper uses the miniscope to image cancer cells in the vasculature, a quite different goal.

freely moving dual photostimulation and imaging

Last month I talked about a paper from the Tank lab detailing dual photostimulation and imaging (DPSI) of neural activity (see dual photo-stimulation and imaging). There is now a new paper from Valentina Emiliani’s group that demonstrates DPSI with freely moving animals. They employ several tricks to get around their non-optimal choice of using ChR2-tdTomato and GCaMP5-G to stimulate and image (they have overlapping excitation spectra). They demonstrate that with some modifications to the imaging, they can reduce background excitation. However, their videos demonstrating dual excitation and imaging are not the most convincing. However, at least they have videos, seems like some imaging papers don’t include any which always makes one a bit suspicious.

wireless sensors for monkeys

Wireless recording has been reported previously and new paper adds another wireless recording device to the list. Whether we will see a similar trend happen with miniature microscopes wait to be seen, their bandwidth requirements are much higher.

go to post.

2014.12.13 biafra ahanonu 2015-01-05T00:00:00+01:00 2015-01-05T00:00:00+01:00

2014.12.13 [link]

using microprisms to access hard to reach brain areas


Figure 44: Skematics for accessing mPFC and MEC using microprisms.

Some regions of the brain are difficult to access for optical imaging due to their awkward orientation or the location of obstacles (e.g. blood vessels) that make it difficult to access without causing serious damage to overlying tissue or structures. The Tank lab has released a paper outlining a method that uses microprisms to help image the prefrontal cortex (involved in emotion, executive processing, and other higher-level cognitive functions) and medial entorhinal cortex (associated with grid cells that help with navigation of environments). Each structure is either occluded by blood vessels or located at an awkward angle, respectively. This build off previous work, such as a the Levene lab’s paper from last year (see below).

go to post.

2014.12.03 biafra ahanonu 2015-01-05T00:00:00+01:00 2015-01-05T00:00:00+01:00

2014.12.03 [link]


Figure 45:

some interesting papers/articles

managing the paper overload

There are a myriad of ways to discover research articles: Pubmed (can create personalized rss feeds from searches), Google Scholar (great for finding alternative versions of articles), Web of Knowledge, school library websites, etc. However, there is still a vast trove of papers being published that can be hard to keep up with. Sciencescape hopes to alleviate that problem by providing personalized updates on specific areas of science. Time will tell if this will help people discover and digest research articles better.

go to post.

2014.11.30 biafra ahanonu 2014-12-01T00:00:00+01:00 2014-12-01T00:00:00+01:00

2014.11.30 [link]

neuroscience technologies

There has been, and continues to be, an explosion in technologies with neuroscience applications. Keeping track of them all can be daunting at times, especially for someone entering the field. I’ve decided to begin assembling a list of neurotechnologies into a living document. Very early draft form is below on Google Doc, still deciding on the best way format and share such a resource.

go to post.

2014.11.29 biafra ahanonu 2014-11-30T00:00:00+01:00 2014-11-30T00:00:00+01:00

2014.11.29 [link]


Figure 46: Doric flyer from SfN.

doric miniature microscope

Doric now has its own line of miniature microscopes to compete with Inscopix and NeuralMapper (see my previous post on the subject).

go to post.

2014.11.28 biafra ahanonu 2014-11-30T00:00:00+01:00 2014-11-30T00:00:00+01:00

2014.11.28 [link]

rebuttal of the trillion odor hypothesis


Figure 47: Two mixtures can share a different number of individual odors. This can be used to probe the number of odors a system can distinguish.[? ]

Markus Meister has a rebuttal to the recent Science paper that uses a simple assay and projections to predict that humans can discriminate over 1 trillion odors. A brief reading of the paper and rebuttal gives the impression that there is a subtle distinction between how one wants to define ‘discriminate’ in terms of the dimensionality of the sensory systems detectors and how the animal actually perceives the sensations.


Figure 48: The rebuttal points out that there is a distinction between sensory precepts being different from their nearest neighbors and being truly unique throughout the precept space.[? ]

For example, subjects report that many complex odor mixtures (composed of  30 individual odor components) actually smell alike.[? ] So while a pairwise comparison between many complex mixtures might show discrimination, it does not demonstrate that we actually can perceive all of them differently (in the sense that we could describe the aroma or associate it with a distinct event/object), only that we can tell the difference between mixtures that are nearby in odor-space. This is a subtle point and Markus does a good job teasing apart where the paper appears to have gone off-track and also suggests how one could design an experiment to properly test the question (how many odors can humans detect). Give both papers a read, it’s well worth the effort.

3D Light Field Microscopy

The below is a useful resource for those wanted to get up to speed on light field microscopy.

go to post.

2014.11.23 biafra ahanonu 2014-11-30T00:00:00+01:00 2014-11-30T00:00:00+01:00

2014.11.23 [link]


Figure 49: Setup for dual photostimulation and imaging of individual neurons.[? ]

dual photo-stimulation and imaging

The Tank lab published a paper showing one way of implementing a set of experiments I had been thinking about for awhile and noted in an older post should allow us to better probe neural activity rather than bulk population optogenetics, see physiological optogenetics.

In brief, the paper demonstrate that one can image neural activity (using GCaMP3, an indicator of calcium activity, which is a surrogate for neural activity) and simultaneously stimulate the same neurons individually (using C1V1, a protein that excites cells when red light is shone on them). This idea of measuring neural activity then stimulating those same cells has been around, and several groups have already demonstrated single-cell optogenetic excitation using patterned illumination and other techniques (see below for several examples/reviews).


Figure 50: A classic experimental setup to directly test the necessity and sufficiency of neuronal activity observed during a behavior.

The paper does a good job showing the technique’s robustness and some limits, but more details on the other optogenetic protein and calcium sensor pairs they claimed to have tested would have been appreciated. The supplemental is rather weak and a video demonstrating the dual stimulating/imaging should have been included (not sure why this isn’t already a requirement for published imaging experiments).

Part of the study looks at using low-power stimulation to perturb network activity and reveal sub-threshold activity of individual cells, which should lead to some interesting applications that build off of Albert Lee's work looking at silent place cells: if combined with recent advances in voltage imaging, this could allow the neuroscience community to address fundamental questions relating to how the intrinsic dynamics of neurons leads to their recruitment during behavior.

This paper, if the technique holds, is going to be a significant new technique in the field and has potential applications beyond neuroscience.1

go to post.

2014.11.22 biafra ahanonu 2014-11-30T00:00:00+01:00 2014-11-30T00:00:00+01:00

2014.11.22 [link]


Figure 51: Richard Feynman during one of his well known lectures on physics.

presenting scientific research

Recently had to give a grad seminar and while preparing the talk I spent some time trying to reduce the complexity of the presentation. This involved removing whole sections of introduction and data analysis while adding in several ‘big picture’ slides to put the study of network dynamics into context, one in terms of a basic scientific question at the beginning and the other in terms of applications of the research. However, the presentation still contained too much nitty-gritty data analysis and I could tell people were getting lost at points. Decided it was time to dig through the presentation resources again, below are a couple that may be of use.


Figure 52: The end goal is to avoid producing slides like this (source).

go to post.

2014.11.20 biafra ahanonu 2014-11-20T00:00:00+01:00 2014-11-20T00:00:00+01:00

2014.11.20 [link]


Figure 53: Imaging setup from recent paper proposing a large field-of-view two-photon microscope.

simultaneous imaging in multiple brain regions

Several papers came out in August showing large field of view imaging (see large scale imaging and other papers). A new paper on the bioRxiv demonstrates a wide field-of-view two-photon microscope using a temporally multiplexed input beam pattern (offset one of the beams by 6.25ns so you can temporally separate the photons coming into a single PMT). However, this system is currently quite slow, taking more than a second to acquire an entire frame. On the other hand, Jerome Lecoq, a postdoc in Mark Schnitzer’s lab (where I am currently working), has a paper out in Nature Neuroscience where they are looking at imaging two areas using microendoscopes. The advantage of this over a large FOV two-photon setup is that you can use GRIN lenses to image two distant, potentially deep-lying brain areas (i.e. subcortical) at the same time. Papers below.

go to post.

2014.10.06 biafra ahanonu 2014-10-07T00:00:00+02:00 2014-10-07T00:00:00+02:00

2014.10.06 [link]

Nobel prize goes to identifying brain regions involved in spatial navigation

May-Britt Moser and Edvard I. Moser have received the Nobel Prize in Physiology or Medicine along with John OKeefe (NYTimes article here) for helping describe how the brain forms internal representations of place. They have many great reviews on the subject over the years, but one I was reading recently before the prizes were announced was Place Cells, Grid Cells, and the Brain's Spatial Representation System in the always excellent Annual Review of Neuroscience.

GCaMP6 reporter mice

While I still don’t quite trust a mouse that has an extra calcium buffer expressed everywhere during development and adult life, there is now a Thy1 reporter mouse for those who’d rather avoid injecting GCaMP via a virus into your brain region of interest before imaging.

go to post.

2014.09.30 biafra ahanonu 2014-10-01T00:00:00+02:00 2014-10-01T00:00:00+02:00

2014.09.30 [link]

NIH BRAIN awards

President Obama’s BRAIN initiative has entered its second phase. Earlier this summer the BRAIN working group released the BRAIN 2025: A Scientific Vision report, which covered the scientific and other rationales for areas where the BRAIN initiative should focus funding. In addition, the NIH announced today the initial $46 million for various project, see NIH BRAIN awards. There are brief descriptions on each of the projects that obtained funding along with more detailed descriptions on NIH's RePORT site. More info in the links below:

go to post.

2014.08.22 biafra ahanonu 2014-08-28T00:00:00+02:00 2014-08-28T00:00:00+02:00

2014.08.22 [link]

large scale imaging and other papers

Large scale imaging of multiple brain areas at a cellular level is the next stage for systems neuroscience. Multiple brain regions are known to be involved in the same behavior and while imaging/recording from each individually has lead to many discoveries on how the brain encodes sensory stimuli and performs computations before performing behaviors, there is a loss of information when doing so. Hence, there is a trend toward recording from multiple brain areas or regions in the same brain area, which should allow us to discern how information flows throughout the brain via correlations in network activity between these areas. Below are a couple interesting papers from this week, including one that does some large-scale imaging.

go to post.

2014.08.17 biafra ahanonu 2014-08-17T00:00:00+02:00 2014-08-17T00:00:00+02:00

2014.08.17 [link]

limits of neuroscience

Came across an interesting New York Times opinion piece (The Trouble With Brain Science) by Gary Marcus. He makes a good point that aspects of both the Brain Initiative (USA) and Human Brain Project (Europe) are flawed in that they don’t seem to address fundamental questions and push for technological over conceptual advances. However, he fails to point readers (or only mentions pop-science books) toward other articles/books that discuss on a more technical, or layman’s, way where neuroscience is heading and what a theory of neuroscience or biology would look like.

Pointing toward physics and saying how they’ve got it figured out is easy, but it doesn’t help in conceptualizing what biologists’ end goals should be. Should we converge on a statistical (thermodynamics), first-principles/observational (classical physics), or some combination of the two? e.g. if i am a big pharma company, would my goals for understanding biology be different than a theoretical systems biologist?

In one case you might just want to do enough experiments that you have large look-up tables for every possible combination of genes and their resulting correlation to cellular action/animal behavior while having secondary databases documenting effects of drugs on these correlations. You won’t need to necessarily know down to the atom exactly how protein A interacts with proteins B-Z, only that protein A has this weak effect and we can correct it (in the case of diseases) with drug A. On the other hand, you might want to build computational models at some level of abstraction—molecules, cells, organ systems, etc.—to simulate the body and thus be able to apply perturbations artificially and see if they match experimental results, validating the underlying models to a degree. These would involve very different technologies and approaches.

Anyways, it seems that Aping Mankind, A Skeptic's Guide to the Mind, and others are good starting points for those wishing to get a picture of the limits of neuroscience and get an alternative view to the hype machine surrounding the USA and Europe’s recent pushes in neuroscience.

on interstellar travel

This entry is a bit farther afield, but was inspired by a chat with a friend about interstellar travel (partially stemming from a conversation about the excellent the Martian) and how crews would respond if they knew that they would be a stop-gap on a multi-generational mission across the stars. In that sense, it is psychology-related and thus neuroscience related. Bam. Relevant.

So the basic premise is this. Assuming we don’t discover faster-than-light travel and are only ever able to attain several percents the speed of light, it will probably take multi-generational missions to reach other planets. Several questions arise when trying to plan such a mission:

  • can people survive the radiation for that long/will offspring be viable?
  • what psychological problems will evolve for people who know they will die in the middle of space (and can we simulate/study such conditions on Earth)?
  • can you instill the same discipline in second, third, and other generations to prevent politics and other issues from distrupting the mission?
  • how do you deal with inbreeding and are all possible matches pre-planned beforehand to reduce this?
  • are there fundamental differences in space travel beyond the heliosphere?
  • how do you adjust course once inside another star system given that most planets are only detected by the wobble of their host star and their location isn’t exact?
  • on more ethical grounds, can we justify terminating or corralling extraterrestrial species if it allows us to properly terraform a planet?
  • more to the point, does Manifest Destiny apply to the cosmos (especially when the fate of our species is at stake)? Or should we follow a doctrine similar to the Prime Directive from Star Trek of non-interference?
  • what mode(s) of travel are most reasonable and do you need multi-stage systems to assemble the necessary supplies in a cost-efficient manner?

These are just some of the questions that first came to mind and there are many more. It seemed prudent to do a little background research and see what the status was of the field. I’ve compiled a couple books/articles that seem useful and will add to the list as i find more. I might look into contacting those working in this area to get a better idea of the state-of-the-art (especially given Inspiration Mars Foundation and other concrete plans to start testing long-term space flight, ignoring whether these private enterprises will succeed).

go to post.

2014.08.16 biafra ahanonu 2014-08-16T00:00:00+02:00 2014-08-16T00:00:00+02:00

2014.08.16 [link]

limits of computation


Figure 54:

Igor Markov has a great review on the fundamental limits to computation. This article reminds me of an older Science article about desalination (The Future of Seawater Desalination: Energy, Technology, and the Environment) that also touches on the fundamental limits in that area by focusing on thermodynamic properties of the process. It is nice that this current paper on computation also talks about the flexibility of what we know regarding limits, thus giving hope for more future advances.

designing brain-based computers


Figure 55:

Von Neumann architecture has been the prevailing method for computer design. However, it runs into issues when trying to do massive parallel computing, something the brain is optimized for. A new paper in science explore design of an integrated circuit based on brain architecture. This seems to build off older work at Intel and elsewhere on neuromorphic devices.

go to post.

2014.08.13 biafra ahanonu 2014-08-12T00:00:00+02:00 2014-08-12T00:00:00+02:00

2014.08.13 [link]

circuit reviews


Figure 56: microcircuit structure and computations

Found this Current Opinion in Neurobiology issue from 2012 quite interesting. In particular this figure that shows microcircuit structures and their resulting computations is a nice simplification that should maybe be extended to each new circuit characterized in different brain regions. This has been done extensively for the retina, but doing so in the cortex, striatum, and other regions might aid in a more concrete understanding of those circuits and easier integration of knowledge accumulated from those regions into a larger understanding of how they all work together.

go to post.

2014.08.11 biafra ahanonu 2014-08-12T00:00:00+02:00 2014-08-12T00:00:00+02:00

2014.08.11 [link]

list of pharmaceutical company projects

Someone pointed out the following, pretty useful resource:

connectivity tracing


Figure 57: mouse connectome project overview.

The Allen Brain Atlas has been leading the way in collecting, organizing, and sharing systematic data on mouse gene expression and whole brain tracing. Other programs, such as USC’s Mouse Connectome Project, are also helping systematize knowledge in this area. Below are several recent papers that are taking advantage of the explosion in rabies (great overview article) and other retrograde tracers for analyzing regional connectivity and its possible influence on brain function.

go to post.

2014.08.10 biafra ahanonu 2014-08-11T00:00:00+02:00 2014-08-11T00:00:00+02:00

2014.08.10 [link]

recent papers


Figure 58: Model of motor neuron recruitment during different stages of swimming.

Below are a couple interesting papers from the last week. Particularly interesting is the last paper looking at the recruitment of different interneuron and motorneuron populations as zebrafish swim faster, as if the organism has a built-in gearbox.

Some recent papers in Nature:

And always nice to see news ways to address neural coding during chronic pain:

go to post.

2014.07.11 biafra ahanonu 2014-07-11T00:00:00+02:00 2014-07-11T00:00:00+02:00

2014.07.11 [link]

scientific impact


Figure 59: The h-index during academic careers.

In a previous post i talked briefly about quantifying scientific progress. Beyond the literature cited there that explored how resulting scientific productivity compared to funding and review panel scores, there is also another question: how do you quantifying scientific impact? What metrics best account for studies or papers that end up being game changers?

Looking past citations, how do you measure the diffusion of an important paper’s ideas? Especially because they should be useful outside the field in question and thus hard to pinpoint the source when looking through another fields literature. Will we ever have the data to determine where useful ideas pop up? For example, we could have the notes of every scientist and analyze when they read certain articles and when the resulting connecting ideas were formed. And if we did, could we really optimize the discovery process? Those are questions for another day, but for now, the below reading should provide some qualitative and quantitative ideas on how to approach those questions.


Figure 60: Break down in citation predictive power for highly cited papers.[? ]

One interesting finding from Wang, 2013[? ] is that for the most highly cited papers, the correlation between the first couple years citations and end citations breaks down. While this makes intuitive sense, the most astounding findings might be ignored because academia is rather conservative in its adoption of new ideas, it does lead one to wonder in what other cases the measures used to review and fund projects start to break down for the most forward thinking ideas, and how one differentiates those from the plain bad projects.

On another interesting note, the below two papers investigate scientific mobility and impact.

go to post.

2014.06.28 biafra ahanonu 2014-06-28T00:00:00+02:00 2014-06-28T00:00:00+02:00

2014.06.28 [link]



Figure 61: Aspects of the CLARITY workflow.

While the wait is on for useful applications of CLARITY beyond pretty pictures is still on (and the whole endeavor might be superfluous given the new clearing techniques as i've discussed previously), a new paper from the Deisseroth lab seeks to address some of the technique’s problems and provide an overview of the end-to-end needed to get it working (e.g. from clearing to antibody staining to imaging).

Advanced CLARITY for rapid and high-resolution imaging of intact tissues

Cell now publishes letters

We went over Skin β-Endorphin Mediates Addiction to UV Light for a journal club earlier this week. Wasn’t able to give it a look until an hour or so before we started, but apparently this is more similar to a Nature article than a Cell mechanism overload. There are five figures when there should be 2 or 3; there are questionable statistics (i’m still trying to understand the p = 0.9511 in figure 3B); the paper is light on actual mechanism, only starting to hint at p53’s (in keratinocytes) control of POMC and thus β-endorphin; there are no pharmaco-kinetics or indication of where exactly β-endorphin is acting (important given the place preference assay they did with peripheral β-endorphin); there is no supplemental data; and the list goes on. Beyond the headlines that this article is likely to make, it is beyond me why Cell took it upon itself to publish this.

go to post.

2014.06.22 biafra ahanonu 2014-06-22T00:00:00+02:00 2014-06-22T00:00:00+02:00

2014.06.22 [link]

statistics: effect sizes

Most statistics in the biosciences (e.g. cancer biology, neuroscience, etc.) are reported as p-values. There is a long standing problem with p-values, namely that given large sample sizes a small effect can become significant. This has been the case in GWAS studies and is increasingly becoming true in neuroscience and other fields as large datasets become the norm.

Other fields, such as clinical medicine, have taken to reporting the effect size, confidence interval, or other statistics that better reflect the biological, and not just statistical, significance of the finding. Below are two good reviews on the usage of effect sizes and other statistical measures other than p-values that should be used when analyzing data.

spike metrics

While talking with Gaurav Venkataraman about the use and problems of using mutual information to identify cells that are responding to a given stimuli (in the spirit of Shannon's Bandwagon essay), he pointed me toward the work of Jonathan Victor. Below are a couple reviews and papers applying some aspects of spike train metrics to neural coding.


While the legal battle over CRISPR continues to heat up, Feng Zhang and co. have written a review covering the technology and its development that should be useful for anyone wanting to get a good overview of the technique.

Development and Applications of CRISPR-Cas9 for Genome Engineering

physiological optogenetics

A recent paper in Cell (Natural Neural Projection Dynamics Underlying Social Behavior) got me thinking again about how closely optogenetics mimics the physiological activity of the perturbed network. The potentially problems with optogenetics have been discussed (e.g. Optogenetics at a crossroads?) but the real physiological nature of optogenetic activation in different brain regions (e.g. activation of areas involved in a behavior compared to the optogenetic activation) have not been fully characterized. For example, calcium dynamics are different in optogentically induced action potentials (APs) compared to electrically produced ones, see Channelrhodopsin as a tool to investigate synaptic transmission and plasticity.), and are likely to be different from endogenously produced APs.

While most studies have a negative control (e.g. just a fluorophore like eGFP/eYFP that doesn’t respond to light + light during behavior), opto-fMRI studies indicate that blue light does cause alterations in brain activity (fMRI response to blue light delivery in the naïve brain: Implications for combined optogenetic fMRI studies). Whether this has more subtle implications that aren’t addressed by behavioral assays used in most studies remains to be seen.

There are many uses for optogenetics where you don’t need to understand how the brain codes for behavior, so long as you obtain a specific behavioral outcome (especially if it will be used in the clinic, similar to the use of deep brain stimulation along with many small molecule and other types of drugs without knowing their exact mechanisms). Much of the hype around optogentics (Why optogenetics deserves the hype) implicitly uses this as a counter to why optogenetics is not overhyped. Fine. Except in scientific research it is being used to elucidate fundamental circuits in behavior (e.g. Dorsal Raphe Neurons Signal Reward through 5-HT and Glutamate) with an implicit/explicit assumption that these circuits encode for or are involved in the computation of specific behaviors.

Given only a specific subset of cells are active for any given stimulus (watch videos with behavior next to neural data from pretty much any two-photon, single-photon, or electrophysiology studies), something commonly attributed to sparse coding, broad spectrum activation is inherently misleading and no study has systematically activated a random subset of brain regions and assayed the mice in a behavioral battery (e.g. Behavioural battery testing: Evaluation and behavioural outcomes in 8 inbred mouse strains) to see how often significant deviations in behavior are seen given certain stimulation parameters (e.g. increase the power, duration, or frequency of the stimulus until an effect is seen).

There are a host of other fundamental problems with optogenetic experiments. I’m merely pointing out one that seems to be mostly ignored in terms of explicitly stating it as a caveat. Whether neural networks respond in a low dimensional ON/OFF fashion as most optogenetic implicitly assume is most likely false based on imaging and other measurements of endogenous neural activity. New tools to more specifically activate a subset of neurons (this is not referring to genetically defined expression, but precise, activity-based subsets) would likely prove more informative with regards to how the brain encodes environmental stimuli, computes, and drives behaviors than the glut of hyper-/hypo-activation optogenetic experiments. More on that in future posts.

go to post.

2014.06.08 biafra ahanonu 2014-06-08T00:00:00+02:00 2014-06-08T00:00:00+02:00

2014.06.08 [link]

drug industry productivity

I recently posted about a Nature Drug Discovery review article by several AstraZenca scientists. Below are a slew of other articles about drug industry productivity.

Poor Productivity As A Self-inflicted Injury: Who’s Missing The Most Toes, And Why

Who's The Best In Drug Research? 22 Companies Ranked

Pfizer's R\&D Productivity

Munos On Big Companies and Small Ones

R\&D productivity: on the comeback trail

Sheer Economics: How We Got in This Fix

consumer virtual reality

Facebook’s recent acquisition of Oculus VR raised eyebrows (this Wired article gives a great overview of the tech and some history behind its development). A consumer version will be released within a year and it will be interesting to see how it is used within the acedemic community (much like the PS3 was) to help further research. For example, there have been several papers putting rodents into virtual reality (and I worked on a similar project at Janelia for Drosophila), it would be interesting to see if the same mouse-on-a-ball or on a treadmill would work better with a small VR headset on the mouse. This could enable much more precise virtual reality than currently present for the animals.

Some of the possibilities of consumer VR are displayed in the following video: Control VR - The Future of VR and Animation

from the archive: some interesting papers

A couple interesting papers spanning neuroscience to synthetic biology.

Robust multicellular computing using genetically encoded NOR gates and chemical `wires'
Using temperature to analyse temporal dynamics in the songbird motor pathway
Prediction and validation of the distinct dynamics of transient and sustained ERK activation.
Neuronal generation of the leech swimming movement.
Encoding multiple unnatural amino acids via evolution of a quadruplet-decoding ribosome
Monosynaptic restriction of transsynaptic tracing from single, genetically targeted neurons.

from the archive: dendritic computation

Came across an excellent review (it’s a couple years old) on dendritic computation by London and Hausser over at UCL. Started looking into related papers, below are two interesting papers looking at the role of (different parts of) dendrites in neural computation.

Principles Governing the Operation of Synaptic Inhibition in Dendrites

Evidence for a computational distinction between proximal and distal neuronal inhibition.

go to post.

2014.06.03 biafra ahanonu 2014-06-04T00:00:00+02:00 2014-06-04T00:00:00+02:00

2014.06.03 [link]

wireless recording from freely moving monkeys


Figure 62: wireless recording from monkeys

chronic, wireless recordings of large-scale brain activity in freely moving rhesus monkeys

In the Schnitzer lab, we regularly record neural activity from freely moving mice during various operant, Pavlovian, fear conditioning, pain, and other behavioral paradigms. Wireless devices are also available for zebra finches and other animals. Now a recent paper has demonstrated recording of neural activity from a freely moving rhesus monkey. Seems they were able to get 5+ years of recordings from some animals, which is pretty amazing, maybe we can start looking at how aging affects gross network activity.

causal investigation of memory and LTD/LTP


Figure 63: protocol used in the paper, ODI is optically stimulation.

Engineering a memory with LTD and LTP

Many systems neuroscientist will agree that long term depression/potentiation underly some aspects of memory in the nervous system. A new study from the Malinow lab uses optical tools to probe whether fear can be directly manipulated by inducing LTD or LTP. The studies protocol is quite simple and the use of optogenetics over electrical stimulation is unclear (besides optogenetics being in vogue). For example, there have been older studies that look at how inducing LTD/LTP can change learned behavior, such as:

Training-induced and electrically induced potentiation in the neocortex

They also show in the extended data that NMDA receptors are necessary (via MK801 blockade) for the optical conditioned response.

spatial optogenetics


Figure 64: Angled optogenetic stimulation.

Multipoint-Emitting Optical Fibers for Spatially Addressable In Vivo Optogenetics

One of the main problems with most optical fibers used in optogenetic experiments is that they broadly diffuse light from the tips, potentially exciting nearby brain regions not part of the study. This recent paper from the Sabatini lab over at Harvard provides a next step toward spatially directing light coming out of an optical fiber.

go to post.

2014.06.01 biafra ahanonu 2014-06-01T00:00:00+02:00 2014-06-01T00:00:00+02:00

2014.06.01 [link]

updated RSS feed

The rss feed now contains the full text for each entry. I’ll write a proper post on the main site on how I automated this via DOM parenting of each post and use of PHPs DOMDocument and DOMXPath classes.

papers from 9.29

Decided to dig through some old papers from 9.29J, an excellent class taught by Michael Fee at MIT the year I took it. We went over a variety of classic papers in the field and thought I’d share some of them:

The role of dendrites in auditory coincidence detection
Direct visuomotor transformations for reaching
Ion-channel defects and aberrant excitability in myotonia and periodic paralysis
Axonal delay lines for time measurement in the owl's brainstem
Optimizing Sound Features for Cortical Neurons
Multiplicative computation in a visual neuron sensitive to looming
A neuronal network for computing population vectors in the leech
Flexible control of mutual inhibition: a neural model of two-interval discrimination - really enjoyed this paper and ended up doing a presentation for it.
Influence of dendritic structure on firing pattern in model neocortical neurons
Segregation of object and background motion in the retina
Internally generated cell assembly sequences in the rat hippocampus
Auditory spatial receptive fields created by multiplication
Neuronal correlates of parametric working memory in the prefrontal cortex
Neuronal generation of the leech swimming movement
Evidence for a computational distinction between proximal and distal neuronal inhibition

go to post.

2014.05.31 biafra ahanonu 2014-06-01T00:00:00+02:00 2014-06-01T00:00:00+02:00

2014.05.31 [link]

unique author identifiers

It seems journals are finally getting with the times and requiring that authors have unique identifiers. Recently The Journal of Neuroscience ran an editorial, Unique Identifiers for Authors, stating that it will require authors to have an ORCID identifier. This is an alternative to ResearcherID promoted by Thomson Reuters and should be more robust due to its broader support by many organizations—including the NIH, EMBL, Nature, and a variety of other organizations. arXiv has author identifiers, but that system is not very robust or interoperable. Hopefully ORCID will become like the digital object identifier system, which has greatly helped in providing what is in essence pointers for digital documents.

While it borderlines on the absurd that this wasn’t implemented back in the 80s or 90s, given that governments have been giving citizens unique identifiers for decades, this is a welcome step forward. It would be better if they could go back and add unique identifiers to older authors. This could be automated and mistakes fixed via crowd-sourcing.

go to post.

2014.05.29 biafra ahanonu 2014-06-01T00:00:00+02:00 2014-06-01T00:00:00+02:00

2014.05.29 [link]

biological logic and computation

A couple interesting papers about creating bacteria and neuron cultures perform computations. These are in no way recent papers.

boolean logic in bacteria

learning in culture — A transparent organic transistor structure for bidirectional stimulation and recording of primary neurons

connectivity and weighting

patterning circuits in vitro

go to post.

2014.05.27 biafra ahanonu 2014-06-01T00:00:00+02:00 2014-06-01T00:00:00+02:00

2014.05.27 [link]

persistent homology

Was chatting with Ryan Lewis (who works over at CompTop) about ways to represent similarity in neuronal firing patterns in response to stimuli. The main thrust of this was the following: supposed in response to each stimuli, a stochastic number of neurons is activated from the entire population. Thus, if one was to use mutual information or Z-scores to help determine which neurons encoded for the stimuli, both measures would end up showing very little correlation between stimuli and neural activity, even if by eye this is obviously not the case. I’ve been interested in whether a higher dimensional representation of the neural activity can be encoded in the pattern of activation and that the evolution of the patterned activation over time can somehow be mapped between each stimuli presentation. This is something like cell assemblies as proposed by Buzaki.

However, while taking a stats class last year, I happened upon temporal exponential random graphical models (tERGMs). These seemed like a reasonable way of representing the evolving neural activity and getting some idea of the associations between cells during behavior.

Going back to my discussion with Ryan, I’m always open to new ways of quantifying neural activity. In this case, persistent homology seemed like an excellent method assuming we could find some specific way to topologically map the neural patterns into generalizable classes, similar to what was done in the below paper:

Encoding Through Patterns: Regression Tree–Based Neuronal Population Models

And more specifically, it appears that the topological analysis can be performed at different scales, in essence allowing one to determine at what spatial scales the stimuli are being encoded in the neuronal ensemble. Before thinking to deeply about the applications of persistent homology, I asked Ryan to send over a reading list to get up to speed on the subject. See below.

Topological Pattern Recognition for Point Cloud Data
Topology for computing
Three Examples Of Applied and Computational Homology
Simplicial Models and Topological Inference in Biological Systems

go to post.

2014.05.25 biafra ahanonu 2014-06-01T00:00:00+02:00 2014-06-01T00:00:00+02:00

2014.05.25 [link]

DREADDing feeding


Figure 65:

Chemogenetic Synaptic Silencing of Neural Circuits Localizes a Hypothalamus→Midbrain Pathway for Feeding Behavior

Always nice to see a paper use DREADDs in an effective manner to dissect a neural circuit, this time the ever fascinating and elusive one involved in feeding behavior. Further, this paper introduces hM4Dnrxn, a modified DREADD that only silences terminals. This allows specific silencing of a regions projection targets while leaving the regions integration and signaling intact. This is useful for circuit tracing, as the effects of having an added G-coupled receptor acting at the soma can have greater unintended consequences compared to one localized at the terminals.

go to post.

2014.05.24 biafra ahanonu 2014-06-01T00:00:00+02:00 2014-06-01T00:00:00+02:00

2014.05.24 [link]

whole animal 3D imaging

There was a recent paper from Janelia using light-sheet microscopy to image an entire zebrafish at 0.8Hz (see Ahrens, 2013). Another paper from Janelia examined zebrafish embryogenesis using light-sheet microscopy as well (Keller, 2013) while Eric Betzig (also at Janelia) has recently published on methods to image large volumes rapidly (Wang, 2014). Vaziri and Boyden have now published a similar application of light-sheet microscopy, only this time they are imaging both C. elegans and larval zebrafish.

Simultaneous whole-animal 3D imaging of neuronal activity using light-field microscopy

That recent paper reminded me of another paper from last year. Bargmann and Albrecht over at Rockefeller used bright-field microscopy to analyze the neuronal activity of around 20 C. elegans simultaneously (Larsch, 2013). My lab is pursuing a similar project, only this one will image the brains of flies (e.g. D. melanogaster). A similar project was published that used genetically driven expression of channelrhodopsin in Drosophila to help tease apart neural correlates of behavior (see Vogelstein, 2014). Whether large-scale projects like these will yield biological conclusions remains to be seen. However, it seems like they should follow the big-science as a map, e.g. like the Human Genome project was.

optical clearing

Application of tissue clearing to imaging of optic nerve (title was intended to be a pun...)

Application of Tissue Clearing and Light Sheet Fluorescence Microscopy to Assess Optic Nerve Regeneration in Unsectioned Tissues

go to post.

2014.05.21 biafra ahanonu 2014-05-21T00:00:00+02:00 2014-05-21T00:00:00+02:00

2014.05.21 [link]

AstraZeneca’s pipeline

While Pfizer’s attempts to buyout AstraZeneca continue to falter, a recent review in Nature Reviews Drug Discovery by several senior scientist at AstraZeneca might point toward why a PfizerAstraZeneca might not have been the best plan from a pipeline perspective anyway.

Lessons learned from the fate of AstraZeneca's drug pipeline: a five-dimensional framework

go to post.

2014.05.12 biafra ahanonu 2014-05-13T00:00:00+02:00 2014-05-13T00:00:00+02:00

2014.05.12 [link]

computational neuroscience papers, part 1

A compilation of some classic (computational) neuroscience papers. This is part I of the series, more to follow!

SB Laughlin. A simple coding procedure enhances a neuron's information capacity. Z Naturforsch, 36c:910-912, 1981.

Lettvin, J.Y., et al.. What the Frog's Eye Tells the Frog's Brain, Proc. Inst. Radio Engr. 47:1940-1951, 1959.

Ballard, DH, Cortical connections and parallel processing: Structure and function, in Vision, in Brain and cooperative computation, pp 563-621, 1987, Arbib, MA and Hanson AR (Eds).

Y Weiss, et al.. Motion illusions as optimal percepts. Nature Neuroscience, 5(6):598 604, 2002.

BA Olshausen and DJ Field. Sparse coding of sensory inputs. Current Opinion in Neurobiology, 14:481 487, 2004.

T Poggio, V Torre, and C Koch. Computational vision and regularization theory. Nature, 317:314 319, 1985.

Girosi, Federico, Michael Jones, and Tomaso Poggio. Regularization theory and neural networks architectures. Neural computation 7.2 (1995): 219-269.

AA Stocker and EP Simoncelli. Noise characteristics and prior expectations in human visual speed perception. Nature Neuroscience, 9(4):578 585, 2006.

Marr, D., and T. Poggio. Cooperative Computation of Stereo Disparity. Science, 194, 283-287, 1976.

Rumelhart, D. E., et al.. Learning representations by back-propagating errors. Nature, 323, 533–536.

Hinton, Geoffrey E., and Steven J. Nowlan. How learning can guide evolution. Complex systems 1.3 (1987): 495-502.

Hinton, G. E. and Plaut, D. C. Using fast weights to deblur old memories. Proceedings of the Ninth Annual Conference of the Cognitive Science Society, Seattle, WA

Becker, S. and Hinton, G. E. A self-organizing neural network that discovers surfaces in random-dot stereograms. Nature, 355:6356, 161-163

Ackley, David H., Geoffrey E. Hinton, and Terrence J. Sejnowski. A learning algorithm for Boltzmann machines. Cognitive science 9.1 (1985): 147-169.

Durbin, Richard, and David Willshaw. An analogue approach to the travelling salesman problem using an elastic net method. Nature 326.6114 (1987): 689-691.

Douglas, Rodney J., Kevan AC Martin, and David Whitteridge. A canonical microcircuit for neocortex. Neural computation 1.4 (1989): 480-488.

Swindale, N. V. A model for the formation of orientation columns. Proceedings of the Royal Society of London. Series B. Biological Sciences 215.1199 (1982): 211-230.

Zohary, E, Shadlen, MN and Newsome, WT (1994). Correlated neuronal discharge rate and its implications for psychophysical performance. Nature 370:140-143.

Hopfield, John J. Neural networks and physical systems with emergent collective computational abilities. Proceedings of the national academy of sciences 79.8 (1982): 2554-2558.

Hopfield, John J. Neurons with graded response have collective computational properties like those of two-state neurons. Proceedings of the national academy of sciences 81.10 (1984): 3088-3092.

Hodgkin, Alan L., and Andrew F. Huxley. A quantitative description of membrane current and its application to conduction and excitation in nerve. The Journal of physiology 117.4 (1952): 500. - And might as well read the rest of their 1952 series.

Song, Sen, and Larry F. Abbott. Cortical development and remapping through spike timing-dependent plasticity. Neuron 32.2 (2001): 339-350.

Buonomano, Dean V., and Michael M. Merzenich. Temporal information transformed into a spatial code by a neural network with realistic properties. Science. 1995 Feb 17;267(5200):1028-30.

Wilson, Hugh R., and Jack D. Cowan. Excitatory and inhibitory interactions in localized populations of model neurons. Biophysical journal 12.1 (1972): 1-24.

Pouget A, Deneve S, Duhamel JR (2002) A computational perspective on the neural basis of multisensory spatial representations. Nat Rev Neurosci. 3: 741-747.

Olshausen BA, Field DJ. 1996. Emergence of simple-cell receptive field properties by learning a sparse code for natural images. Nature 381:607-9.

Atick, Joseph J. Could information theory provide an ecological theory of sensory processing?. Network: Computation in neural systems 3.2 (1992): 213-251.

Milo, Ron, et al. Network motifs: simple building blocks of complex networks. Science 298.5594 (2002): 824-827.

Song, Sen, et al. Highly nonrandom features of synaptic connectivity in local cortical circuits. PLoS biology 3.3 (2005): e68.

Chklovskii, Dmitri B., B. W. Mel, and K. Svoboda. Cortical rewiring and information storage. Nature 431.7010 (2004): 782-788.

Traub, Roger D., and Richard Miles. Neuronal networks of the hippocampus. Vol. 777. Cambridge University Press, 1991.

Whittington, Miles A., Roger D. Traub, and John GR Jefferys. Synchronized oscillations in interneuron networks driven by metabotropic glutamate receptor activation. Nature 373.6515 (1995): 612-615.

Pinsky, Paul F., and John Rinzel. Intrinsic and network rhythmogenesis in a reduced Traub model for CA3 neurons. Journal of computational neuroscience 1.1-2 (1994): 39-60.

De Schutter, Erik, and James M. Bower. An active membrane model of the cerebellar Purkinje cell. I. Simulation of current clamps in slice. Journal of neurophysiology 71.1 (1994): 375-400.

Markram, Henry. The blue brain project. Nature Reviews Neuroscience 7.2 (2006): 153-160.

Markram, Henry, Yun Wang, and Misha Tsodyks. Differential signaling via the same axon of neocortical pyramidal neurons. Proceedings of the National Academy of Sciences 95.9 (1998): 5323-5328.

Treves, Alessandro, and Edmund T. Rolls. Computational analysis of the role of the hippocampus in memory. Hippocampus 4.3 (1994): 374-391.

Rolls, Edmund T., Alessandro Treves, and Edmund T. Rolls. Neural networks and brain function. (1998).

London, Michael, and Michael Husser. Dendritic computation. Annu. Rev. Neurosci. 28 (2005): 503-532.

De La Rocha, Jaime, et al. Correlation between neural spike trains increases with firing rate. Nature 448.7155 (2007): 802-806.

Steriade, Mircea, David A. McCormick, and Terrence J. Sejnowski. Thalamocortical oscillations in the sleeping and aroused brain. Science 262.5134 (1993): 679-685.

Bell, Anthony J., and Terrence J. Sejnowski. The `independent components' of natural scenes are edge filters. Vision research 37.23 (1997): 3327-3338.

Churchland, Patricia Smith, and Terrence J. Sejnowski. The computational brain. The MIT press, 1992.

Bhalla, Upinder S., and Ravi Iyengar. Emergent properties of networks of biological signaling pathways. Science 283.5400 (1999): 381-387.

Zador, Anthony, Christof Koch, and Thomas H. Brown. Biophysical model of a Hebbian synapse. Proceedings of the National Academy of Sciences 87.17 (1990): 6718-6722.

Brown, Thomas H., Edward W. Kairiss, and Claude L. Keenan. Hebbian synapses: biophysical mechanisms and algorithms. Annual review of neuroscience 13.1 (1990): 475-511.

Destexhe, Alain, Zachary F. Mainen, and Terrence J. Sejnowski. Synthesis of models for excitable membranes, synaptic transmission and neuromodulation using a common kinetic formalism. Journal of computational neuroscience 1.3 (1994): 195-230.

Liley, A. W., and K. A. K. North. An electrical investigation of effects of repetitive stimulation on mammalian neuromuscular junction. Journal of Neurophysiology 16.5 (1953): 509-527.

Betz, W. J. Depression of transmitter release at the neuromuscular junction of the frog. The Journal of physiology 206.3 (1970): 629.

Abbott, L. F., et al. Synaptic depression and cortical gain control. Science 275.5297 (1997): 221-224.

Tsodyks, Misha V., and Henry Markram. The neural code between neocortical pyramidal neurons depends on neurotransmitter release probability. Proceedings of the National Academy of Sciences 94.2 (1997): 719-723.

Hopfield, John J., and A. V. Herz. Rapid local synchronization of action potentials: Toward computation with coupled integrate-and-fire neurons. Proceedings of the National Academy of Sciences 92.15 (1995): 6655-6662.

Von Der Malsburg, Christoph. The correlation theory of brain function. Springer New York, 1994.

Willshaw, David J., and Christoph Von Der Malsburg. A marker induction mechanism for the establishment of ordered neural mappings: its application to the retinotectal problem. Philosophical Transactions of the Royal Society of London. B, Biological Sciences 287.1021 (1979): 203-243.

Srinivasan, Mandyam V., Simon B. Laughlin, and Andreas Dubs. Predictive coding: a fresh view of inhibition in the retina. Proceedings of the Royal Society of London. Series B. Biological Sciences 216.1205 (1982): 427-459.

go to post.

2014.05.08 biafra ahanonu 2014-05-09T00:00:00+02:00 2014-05-09T00:00:00+02:00

2014.05.08 [link]

new neuroscience tools from janelia


Figure 66: glutamate sensor iGluSnFR

Loren Looger, group leader at Janelia, gave a rather interesting talk today at Stanford. He went over several new technologies being developed at Janelia and the talk had an interesting, choose-your-own adventure style, a nice twist on the normal talk style. In addition, he had executive summaries for parts of the talk that he could not go over in detail. Overall, it was an excellent talk and from the discussion afterwards, it is clear that he understands and is willing to discuss the limitations of the various technologies he’s developing and the ones currently in use.

calcium sensors

Nothing crazy new here. He mentioned that GCaMP7 is in the works and will have a lower baseline background (F0) and better SNR.

neurotransmitter sensors

iGluSnFR (intensity-based glutamate-sensing fluorescent reporter, see Marvin, 2013) can be used as a measure of glutamate release. The kinetics of their most recent version are extremely fast, such that a slow confocal or two-photon might not show any activity because the frame-rate is too slow or averaging is washing the signal out. He mentioned that Lin Tian is working on a split trans-synaptic version, which should allow measurement of synaptic strength between regions, which could be a real boon for people studying plasticity in various paradigms.

The sensor has been used in two recently published papers:
Kainate Receptors Mediate Signaling in Both Transient and Sustained OFF Bipolar Cell Pathways in Mouse Retina
Two-Photon Imaging of Nonlinear Glutamate Release Dynamics at Bipolar Cell Synapses in the Mouse Retina

In addition to a glutamate sensor, he mentioned that they are attempting to develop dopamine, acetylcholine, adenosine, and other neurotransmitter receptors to help people answer questions pretaining to those systems, these new sensors should prove useful to Drosophila researchers where the main neurotransmitter is acetylcholine.[? ] The dopamine sensor reminds me of Alan Jasanoff’s recent dopamine contrast agent paper, Lee, 2014, which uses a sensor based on directed evolution in Shapiro, 2010. While fMRI has several advantages, it is still limited by needing the animal to remain motionless, which is where the sensors could prove useful in combination with the miniature microscopes in Mark Schnitzer’s lab or other imaging technologies in the pipeline.

experiments A brief idea: the DA sensor could prove extremely helpful in deciphering the role of dopamine in both limbic and cortical (especially prefrontal cortex) structures as it relates to modulation of neuronal activity during learning, reward, addiction, and pain. While the cellular effects of the various dopamine receptors are known to a degree, precisely how local dopamine effects neuronal activity has been unavailable. If the dopamine sensor ends up being green, one could take advantage of RCaMP (see Akerboom, 2013) to simultaneous image local dopamine release and neuronal activity.

dual fluorescence and EM imaging

EosFP is an osmium resistant fluorophore that should be useful for helping do tracing and then using scanning electron microscope (SEM) or transmission electron microscopy (TEM) to do more detailed tracing. This can allow for error correction because the sparse labeling via the fluorescent protein can be used to see if there has been a register shift or some other problem. They are trying to make them resistant to EPON, a particular epoxy used to help keep sample rigid during slicing.

imaging fine processes

smFP (spaghetti monster fluorescent proteins, yes he has a sense of humor when naming his technologies) can be used to detect finer details when doing tracing or other studies. They basically have added FLAG, HA, myc, and potentially other protein tags that have been used for decades to allow pull-down in biochemical assays, among a variety of other uses. The tech doesn’t seem revolutionary, but appears to be a useful tool for dissecting neural circuits or things going on in areas with finer synapses and detail. For example, showed CA3 spine thorns, which are much more complex than normal dendritic spines. The data looked quite convincing.

calcium integrators

CaMPARI (calcium modulated photoactivatable ratiometric integrator, also named from the liqueur Campari) is the name given to the new sensor. It works by taking advantage of the fact that some sensors photoconvert to a new state upon being hit by photons while others (like GCaMP calcium sensors) change state when calcium concentration changes.

problems Looger noted several drawbacks of the technique. Because of calcium dynamics, an experimenter can’t use CaMPARI to measure precise spike timing, as spikes before turning on the UV light can lead to calcium transients that leak into the light-on time. In addition, in order to photoswitch, UV light needs to be used, with two obvious problems: this can be damaging to the tissue being manipulated and is amenable to deep brain without use of a light probe or another manipulation.

applications He showed it worked with linear response by stimulating rat hippocampal neurons. Futher, using a flight-response paradigm in zebra fish (e.g. poke them with a pencil to initiate a flight response), they showed that they could get activation in freely moving fish but not those that were immobilized. Further, they partnered with Karel Svoboda to demonstrate the tech could detect direction selectivity of cells in the mouse visual cortex. Lastly, he noted that pilot experiments implied that they could visualize labeled line of the Drosophila olfactory circuit, e.g. olfactory receptor neuron to projection neuron and onward to tertiary and quaternary neurons. Could be quite interesting. Also, he mentioned that it might be possible to use three-photo imaging (this new report gives a nice overview) to do non-invasive, deep imaging of activated neurons in more superficial structures.

experiments I could easily envision several interesting experiments, for example one could look at either cocaine administration or morphine sensitization and for each dose, apply UV light. Then sacrifice the animal and FACS sort the cells then sequence to see whether there are specific cell types that are activated during drug administration (or take advantage of transgenetic Cre animals and td-Tomato reporter lines to just do two-photon alignment). Pain is a field that seems a fair bit behind in terms of systems neuroscience analysis of supraspinal regions. CaMPARI would allow an exploratory look at regions activated during noxious stimuli, pain relief or chronic pain. This might be several order of magnitude more sensitive than TRAP or similar technologies based on immediate early genes. Though, this tech is at the moment more invasive.

designed genetic switches

Apparently DuPont called Looger a decade or so ago to help them develop a system whereby they could turn on specific genes within crops. The incentive was that they had many transgenetic crops or variants that had drought, pest, etc. resistance but they could not establish a line because the F0 was sterile. They initially were looking at the commonly used inducible gene system, tetracycline. However, for obvious health reasons, this could not be put into widespread agriculture use. Thus, they started looking at several pesticides, one of which was chlorsulfuron, which happened to chemically look the nearest to tetracycline. Looger was able to show that they could use guided computational design and directed evolution to make tetracycline repressor become specific to chlorsulfuron (see the patent US 8257956 B2 - Sulfonylurea-responsive repressor proteins). They can induce expression of a reporter protein either use a spray or root absorption.

GWAS 2.0

Looger seemed to focus on the fact that many GWAS studies point toward a general region in the genome or SNPs that are correlated with a particular behavior, but that they just leave it at that, which isn’t very satisfactory from a mechanistic standpoint. He wants to start seeing if there is a systematic way to look at where exactly the SNPs are occurring, enhancers, binding sites, etc. and what the biochemical consequences of these are. Then systematically seeing the changes that occur. He has already started along this track with a recent paper:

Allelic heterogeneity in NCF2 associated with systemic lupus erythematosus (SLE) susceptibility across four ethnic populations

go to post.

2014.05.07 biafra ahanonu 2014-05-07T00:00:00+02:00 2014-05-07T00:00:00+02:00

2014.05.07 [link]

anti-aging brigade


Figure 67: restoration of aging cells by parabiosis

Two recent papers in Science and one in Nature Medicine build on previous work (see Villeda, 2011) looking into the effects of young blood on aging mice. One, from Amy Wagers’s group at Harvard, identifies GDF11 as a possible molecular component of young blood crucial for the anti-aging effects seen, specifically focusing on skeletal muscles. The other, coming out of Lee Rubin’s lab in collaboration with the Wagers lab, also illustrate the role of GDF11, but this time looking at its effect on vasculature and neurogenesis. Lastly, work from the Wyss-Coray lab at Stanford focus on using microarray analysis to identify genes altered by parabiosis of young and old animals. They identify Egr1, an immediate early gene associated with neuronal activity, and Creb as displaying increased expression and phosphorylation status, respectively.

All these studies point toward multiple mechanisms going on leading to the reduction in cognitive and physiological decline. I’ll update the post as I read each article in more detail.

Restoring Systemic GDF11 Levels Reverses Age-Related Dysfunction in Mouse Skeletal Muscle

Vascular and Neurogenic Rejuvenation of the Aging Mouse Brain by Young Systemic Factors

Young blood reverses age-related impairments in cognitive function and synaptic plasticity in mice

go to post.

2014.04.29 biafra ahanonu 2014-04-29T00:00:00+02:00 2014-04-29T00:00:00+02:00

go to post.

2014.04.25 biafra ahanonu 2014-04-26T00:00:00+02:00 2014-04-26T00:00:00+02:00

2014.04.25 [link]

improving optogenetic silencing


Figure 68: bistable optogenetic silencing of neurons

One of the key technological advances in neuroscience over the last decade has been the discovery, development, and optimization of genetically encoded ion channels and pumps, chief amongst them channelrhodopsin, archaerhodopsin, and Halorhodopsin. While channelrhodopsins have become the standard for activation of neurons by light and are quite effective at doing so (and they can also be activated at a range of spectra, see ReaChR and Chronos and Chrimson, the proton and chloride pumps suffer the problem of inefficient usage of photon input to amount of silencing, in many cases it is a one-to-one correspondence of photons to ions moved. Thus, both Disseroth’s and Hegemann’s groups designed inhibitory channelrhodopsins, which should allow improve volumetric inhibition over longer periods of time at lower power (thus reducing photodamage). Further, the bidirectional control of the SwiChR variant in the second paper (see figure 4h) should allow a more efficient, long time-scale modulation of activity.

Conversion of Channelrhodopsin into a Light-Gated Chloride Channel

Structure-Guided Transformation of Channelrhodopsin into a Light-Activated Chloride Channel

using FISSEQ and neuronal barcoding for connectomics

Marblestone, Boyden and the rest have a recent paper out detailing a strategy to do spatial connectomics using the barcoding method outline previously.[? ] This follows a similar strategy I hinted at in the paper I co-authored with others in the cs379c class.[? ] They go over the optical, costs, and other requirements to get such a system to work. Interesting read:

Rosetta Brains: A Strategy for Molecularly-Annotated Connectomics

go to post.

2014.04.24 biafra ahanonu 2014-04-25T00:00:00+02:00 2014-04-25T00:00:00+02:00

2014.04.24 [link]

high throughput screening for neural mechanisms of behavior


Figure 69: experimental design

This is a rather interesting paper coming out of Janelia and fits into its classic mode of combining long-shot science with powerful genetic and technological tools (Science's perspective piece). In the study, they optogenetically manipulated around a thousand different neuron subsets (based on different Drosophila lines) and ended up mapping the resulting behavioral data into 29 distinct phenotypes.

Discovery of Brainwide Neural-Behavioral Maps via Multiscale Unsupervised Structure Learning

go to post.

2014.04.23 biafra ahanonu 2014-04-23T00:00:00+02:00 2014-04-23T00:00:00+02:00

2014.04.23 [link]

grant funding, scientific impact and resource allocation

I plan on writing a longer post that looks into different ways people are attempting to quantify (or Moneyball) scientific progress in terms of what happens during the project and how exactly to determine future success using metrics other that citation. But for now, we will settle with this pair of interesting studies showing a lack of correlation between score received duirng grant review and final ‘impact’ as judged by citations.

Percentile Ranking and Citation Impact of a Large Cohort of NHLBI-Funded Cardiovascular R01 Grants

Productivity Metrics and Peer Review Scores

Interestingly, the last two used the panel review scores as a metric, another possible metric is funding.

Big Science vs. Little Science: How Scientific Impact Scales with Funding

However, that has the problem that more funding can just allow more technologically fancy, but not necessarily biologically informative, studies that are favored in high-impact journals. As the high impact journals necessarily get more citations due to visibility, this confounds funding with better science, which might not necessarily be the case.

An interesting extension of this is whether scientific funding is being efficiently allocated and how one would go about measuring this. Perhaps a more basic question is: because peer review is used to decide which grants get funded, is peer reviewing reliable and efficient way to measure potential scientific impact or value? The following paper

Funding grant proposals for scientific research: retrospective analysis of scores by members of grant review panel

suggest that high variability in review panels assessment of grants can lead to many of grants not receiving funding because reviewer score variation prevents them from ever crossing the funding threshold. It was found that reliability of the panels decision increased when around eleven panel members was used, but how this varies across disciplines still needs to be worked out.

There have been a plethora of papers looking into how efficient the peer review system is, such as the effect of the drive to find ‘important’ research, the desire to find technical flaws (that don’t necessarily impact the scientific merit of the proposal), personal preferences and/or grudges, cheerleader effect (one or two people can sway the group to come to a non-optimal consensus) and various other problems. Many of these are inherit to decision by consensus, though whether this devolves into groupthink is another issue entirely.

This will be a topic for further discussion, but for now here are several papers on peer review:

Sample Size and Precision in NIH Peer Review

Statistical analysis of the National Institutes of Health peer review system

Chance and consensus in peer review

The Predictive Ability of Peer Review of Grant Proposals: The Case of Ecology and the US National Science Foundation

Editorial peer review for improving the quality of reports of biomedical studies

Effects of Editorial Peer Review

Who Reviews the Reviewers? Feasibility of Using a Fictitious Manuscript to Evaluate Peer Reviewer Performance

Effect on the Quality of Peer Review of Blinding Reviewers and Asking Them to Sign Their Reports

Differences in Review Quality and Recommendations for Publication Between Peer Reviewers Suggested by Authors or by Editors

Peer review for improving the quality of grant applications

How reliable is peer review? An examination of operating grant proposals simultaneously submitted to two similar peer review systems

What errors do peer reviewers detect, and does training improve their ability to detect them?

Effects of training on quality of peer review: randomised controlled trial

go to post.

2014.04.22 biafra ahanonu 2014-04-23T00:00:00+02:00 2014-04-23T00:00:00+02:00

2014.04.22 [link]

CUBIC: another clearing technique


Figure 70: new CUBIC clearing technique outline.

Whole-Brain Imaging with Single-Cell Resolution Using Chemical Cocktails and Computational Analysis

Check out the supplemental info for movies of the 3D reconstructions.

Add it to the list of recent clearing techniques that include: SeeDB, ClearT, and CLARITY.

go to post.

2014.04.20 biafra ahanonu 2014-04-22T00:00:00+02:00 2014-04-22T00:00:00+02:00

2014.04.20 [link]

SwissTech Convention Center


Figure 71: an excuse to have a conference in switzerland

Every once in awhile you see a building that makes you want to find an excuse to visit the city where it is located. The SwissTech Convention Center is one such case. More photos can be found at the flickr page.

More information: The SwissTech Convention Center, a lab for conferences of the future

go to post.

2014.04.19 biafra ahanonu 2014-04-19T00:00:00+02:00 2014-04-19T00:00:00+02:00

2014.04.19 [link]

computational neuroscience resources

The field of computational neuroscience is rather vast, from biophysical models of ion channel opening and closing to network models of action selection in the basal ganglia. Trying to find relevant papers is often hard. Below are a couple resources that may prove useful. I will add more as they come along.

go to post.

2014.04.13 biafra ahanonu 2014-04-19T00:00:00+02:00 2014-04-19T00:00:00+02:00

2014.04.13 [link]

photoacoustic imaging


Figure 72: setup for photoacoustic imaging.[? ]

Mouse brain imaging using photoacoustic computed tomography

go to post.

2014.04.06 biafra ahanonu 2014-04-14T00:00:00+02:00 2014-04-14T00:00:00+02:00

2014.04.06 [link]

calcium imaging cell detection techniques


Figure 73: performance of NMF vs. PCA-ICA.

In our lab we routinely need to identify cells in calcium imaging (normally with GCamP variants) done in either a two-photon setup or with miniature microscopes. The current standard is to use the PCA-ICA method developed in the lab.[? ] This works quite well, but there are limitations to the method (and here are a few): namely algorithmically it is implemented with a random seed, hence the results are not always consistent run-to-run; for best results is requires a priori knowledge of the number of signals (i.e. cells) that need to be extracted; and there is a massive problem with cross-talking in each signal from signals that are spatially nearby.

While a new method for signal extraction is being developed in the lab, i have also been on the search for any new techniques that are coming down the line. This paper, Detecting cells using non-negative matrix factorization on calcium imaging data, attempts to develop a new method to more accurately detect cells that PCA-ICA.

go to post.

2014.04.05 biafra ahanonu 2014-04-14T00:00:00+02:00 2014-04-14T00:00:00+02:00

2014.04.05 [link]

recording from 512 signals


Figure 74: determining connectivity via high density probes.

A new paper on recording from 512 channels in rats. When combined with optogenetics identification of specific cell types, this could prove quite powerful. This could also be used in conjunction with new optogenetic tools that allow excitation of several neuronal populations at once using Boyden’s recently characterized channelrhodopsins Chronos and Crimson[? ]. See link to article below.

Large-scale, high-density (up to 512 channels) recording of local circuits in behaving animals

go to post.

2013.09.14 biafra ahanonu 2013-09-15T00:00:00+02:00 2013-09-15T00:00:00+02:00

2013.09.14 [link]

ethics of science

Some articles related to publication, authorship, and ethical practices in science.

Flows of Research Manuscripts Among Scientific Journals Reveal Hidden Submission Patterns

Impact Factor Distortions

The Misused Impact Factor

Limiting the Impact of the Impact Factor

The mismeasurement of science

Rank injustice

Competition and Careers in Biosciences

Tank injustice and academic promotion

Read before you cite!

Citation opportunity cost of the high impact factor obsession

Quiet debut for the double helix

The politics of publication

Addressing the Nation’s Changing Needs for Biomedical and Behavioral Scientists

from the archive

A gene complex controlling segmentation in Drosophila—classic review article by Lewis.

Shank3 mutant mice display autistic-like behaviours and striatal dysfunction—nice paper characterizing Shank3 mice. Related: Mutations in the gene encoding the synaptic scaffolding protein SHANK3 are associated with autism spectrum disorders.

Ecosystem consequences of bird declines

random topic

Latent Dirichlet allocation

Syntactic Ngrams over Time

go to post.

2013.09.10 biafra ahanonu 2013-09-10T00:00:00+02:00 2013-09-10T00:00:00+02:00

2013.09.10 [link]

The focus today is on papers!

brain initiative

The 302 neurons and 7,000 connections that make up the nervous system of the roundworm Caenorhabditis elegans were mapped in the 1970s and 80s. More than two decades later, little is understood about how the worm’s nervous system produces complex behaviours.Douglas Fields

Neuroscience: Map the other brain—talks about the need to also study glia. See Ben Barres for more on this topic.

The benefits of brain mapping

Neuroscience: Brain projects need stronger foundation

just plain cool

Time crystals and the Nature article: Can matter cycle through shapes eternally?.

Soft tissue preservation in a fossil marine lizard with a bilobed tail fin

Synthesizing cognition in neuromorphic electronic systems—see neuromorphic engineering for more. There is an extensive literature on the subject.[? ? ? ]


DNA methylation regulates associative reward learning—pretty good paper looking at the neurobiology of reward learning, in this case the role of DNA methylation, which is known to alter gene expression.

Recovery from slow inactivation in K+ channels is controlled by water molecules

Topoisomerases facilitate transcription of long genes linked to autism—the accompanying layman’s explanation: Autism: A long genetic explanation.

Bacteria activate sensory neurons that modulate pain and inflammation—the layman’s article: Bacteria get on your nerves. This is a super cool paper looking at how a particular bacteria, S. aureus can modify the state of sensory neurons.

Self-propagation of pathogenic protein aggregates in neurodegenerative diseases

Video game training enhances cognitive control in older adults—the media loves these types of studies.

Distinct Representations of Cognitive and Motivational Signals in Midbrain Dopamine Neurons

Integration of GABAergic Interneurons into Cortical Cell Assemblies: Lessons from Embryos and Adults

Temporally Precise Cell-Specific Coherence Develops in Corticostriatal Networks during Learning

Distinct Basal Ganglia Circuits Controlling Behaviors Guided by Flexible and Stable Values

Topographic Representation of Numerosity in the Human Parietal Cortex

A Causative Link Between Inner Ear Defects and Long-Term Striatal Dysfunction

Assignment of Model Amygdala Neurons to the Fear Memory Trace Depends on Competitive Synaptic Interactions

Neural Representation of a Target Auditory Memory in a Cortico-Basal Ganglia Pathway

Reward Learning Requires Activity of Matrix Metalloproteinase-9 in the Central Amygdala

Advances in the pharmacological treatment of Parkinson's disease: targeting neurotransmitter systems

Neuronal circuits that regulate feeding behavior and metabolism


Because i secretly still wish to be working with this amazing model organism.

Odor Discrimination in Drosophila: From Neural Population Codes to Behavior

Neuroendocrine Control of Drosophila Larval Light Preference

Shocking Revelations and Saccharin Sweetness in the Study of Drosophila Olfactory Memory


And just when you thought the CRISPR fun had gone cold!

One-Step Generation of Mice Carrying Reporter and Conditional Alleles by CRISPR/Cas-Mediated Genome Engineering

Double Nicking by RNA-Guided CRISPR Cas9 for Enhanced Genome Editing Specificity


Evolutionary origins of the avian brain

Computational design of ligand-binding proteins with high affinity and selectivity—and so we edge ever closer to the holy grail of computational design of drugs. This is pretty cool.

Evolution Heresy? Epigenetics Underlies Heritable Plant Traits

Bit-by-bit autophagic removal of parkin-labelled mitochondria

Identification of a splice variant of mouse TRPA1 that regulates TRPA1 activity

Huntington disease arises from a combinatory toxicity of polyglutamine and copper binding

Neuropeptide signaling remodels chemosensory circuit composition in Caenorhabditis elegans

from the archives

Topoisomerase inhibitors unsilence the dormant allele of Ube3a in neurons—a nice study that used a small molecule screen to study UBE3A, a gene that in neurons is exclusively expressed by the maternal chromosome.

go to post.

2013.09.05 biafra ahanonu 2013-09-05T00:00:00+02:00 2013-09-05T00:00:00+02:00

go to post.

2013.09.04 biafra ahanonu 2013-09-04T00:00:00+02:00 2013-09-04T00:00:00+02:00

go to post.

2013.09.03 biafra ahanonu 2013-09-03T00:00:00+02:00 2013-09-03T00:00:00+02:00

2013.09.03 [link]

tech articles

ReaChR: a red-shifted variant of channelrhodopsin enables deep transcranial optogenetic excitation—this will be useful both for deep brain stimulation and possibly allowing more simultaneous optogenetic experiments.

A database of Caenorhabditis elegans behavioral phenotypes—databases are always welcome. An equivalent for mice can be found at JAX.

Functional labeling of neurons and their projections using the synthetic activity–dependent promoter E-SARE

Engineering of weak helper interactions for high-efficiency FRET probes

Plasmonic gold mushroom arrays with refractive index sensing figures of merit approaching the theoretical limit

Imaging electrical activity of neurons with metamaterial nanosensors—actual implementation and experimental results would have made this paper much better.

The COMBREX Project: Design, Methodology, and Initial Results—more projects like this are needed to help bridge the experimental/computational drive.

A brief account of nanoparticle contrast agents for photoacoustic imagingphotoacoustic imaging could be big going forward, especially given newer applications coming down the pipeline.

Optical fibers for high-resolution in vivo microendoscopic fluorescence imaging

from the archives

Some articles by Cori Bargmann, co-chair on the BRAIN initiative.

How (not) to get a job...

Graduate school: the movie

Decisions, decisions

go to post.

2013.09.02 biafra ahanonu 2013-09-02T00:00:00+02:00 2013-09-02T00:00:00+02:00

2013.09.02 [link]


Figure 76: USA federal R&D spending.

ASBMB report

ASBMB has released a useful report looking at federal budget cuts to research programs: Unlimited Potential, Vanishing Opportunity

go to post.

2013.09.01 biafra ahanonu 2013-09-01T00:00:00+02:00 2013-09-01T00:00:00+02:00

2013.09.01 [link]


Some interesting neuro-related articles from around the web.

Serotonin and the Neuropeptide PDF Initiate and Extend Opposing Behavioral States in C. elegans—Cori gave a speech on this several months ago at Stanford, nice to see the final report.

Calling the next generation of affinity reagents

Ethical reproducibility: towards transparent reporting in biomedical research

Emergent Properties of the Optic Tectum Revealed by Population Analysis of Direction and Orientation Selectivity

Nuclear calcium signaling in the regulation of brain function

Simultaneous PET-MRI reveals brain function in activated and resting state on metabolic, hemodynamic and multiple temporal scales

Transplantation reveals regional differences in oligodendrocyte differentiation in the adult brain

Population Coding and the Labeling Problem: Extrinsic Versus Intrinsic Representations

Evidence for Hubs in Human Functional Brain Networks—reminds me a bit of Why Do Hubs Tend to Be Essential in Protein Networks?

An Interactive Resource to Identify Cancer Genetic and Lineage Dependencies Targeted by Small Molecules

Equating information-theoretic and likelihood-based methods for neural dimensionality reduction

Sparse Coding Models Can Exhibit Decreasing Sparseness while Learning Sparse Codes for Natural Images—a nice revisiting of the sparse coding hypothesis.

Causes and Consequences of Hyperexcitation in Central Clock Neurons

Characteristic Effects of Stochastic Oscillatory Forcing on Neural Firing: Analytical Theory and Comparison to Paddlefish Electroreceptor Data

Information and Efficiency in the Nervous System--A Synthesis—there is a whole literature on efficiency in the nervous system. Perhaps the best come from David Attwell and co.

Autonomous molecular cascades for evaluation of cell surfaces

Energy-efficient encoding by shifting spikes in neocortical neurons

Suppressing aberrant GluN3A expression rescues synaptic and behavioral impairments in Huntington's disease models

from the archives

The office of the future: a unified approach to image-based modeling and spatially immersive displays

The Programmer's Apprentice Project: A Research Overview - thanks to Tom Dean for pointing this article out.

Transgenic and knockout databases: Behavioral profiles of mouse mutants—see JAX for current database.

just plain cool

Accelerated chemistry in the reaction between the hydroxyl radical and methanol at interstellar temperatures facilitated by tunnelling—see the corresponding layman's report.

A comprehensive multiscale framework for simulating optogenetics in the heart

The Extraordinary Evolutionary History of the Reticuloendotheliosis Viruses

Nuclear Lamin-A Scales with Tissue Stiffness and Enhances Matrix-Directed Differentiation


Some websites that are relevant in terms of expanding the repitoire of resources avaliable to researchers in neuroscience (and biology in general).

DBpedia and Freebase — these could provide a basis for new website that provides semantic curration of research results.

Schema—a basis for semantic mark-up of scientific sites.

Comparison of Linux Development Boards—these boards can be used for a variety of tasks that can have applications in prototyping experiments.

go to post.

2013.08.29 biafra ahanonu 2013-08-29T00:00:00+02:00 2013-08-29T00:00:00+02:00

go to post.

2013.08.28 biafra ahanonu 2013-08-28T00:00:00+02:00 2013-08-28T00:00:00+02:00

go to post.

2013.08.25 biafra ahanonu 2013-08-25T00:00:00+02:00 2013-08-25T00:00:00+02:00

2013.08.25 [link]


Here is a sample of some interesting articles from around the web concerning neuro research.

Mapping Neuronal Diversity One Cell at a Time - looking into the past literature (The neuron classification problem) gives a sense of how hard this problem is of identifying neuron types.

Probabilistic brains: knowns and unknowns

Neuroscience thinks big (and collaboratively)

Balanced cortical microcircuitry for maintaining information in working memory - reminds me a bit of Machens’ 2005 paper (Flexible Control of Mutual Inhibition: A Neural Model of Two-Interval Discrimination).

Lineage-specific laminar organization of cortical GABAergic interneurons

Wnt/Dkk Negative Feedback RegulatesSensory Organ Size in Zebrafish - had to do a report looking at how to identify what molecules are involved in determining organ size, this is a good study looking into that.

Food Restriction Increases Glutamate Receptor-Mediated Burst Firing of Dopamine Neurons

Dopaminergic Control of Long-Term Depression/Long-Term Potentiation Threshold in Prefrontal Cortex

from the archives

A couple older articles related to the above:

Probing perceptual decisions in rodents

Evaluation of Excess Significance Bias in Animal Studies of Neurological Diseases - nice study looking at the bias in reporting in animal studies.

go to post.

2013.08.22 biafra ahanonu 2013-08-22T00:00:00+02:00 2013-08-22T00:00:00+02:00

2013.08.22 [link]

towards an integrated science

Reading papers is both exhilarating, dull, challenging, and ultimately inefficient. While the specifics of a paper might be important for a project—you want to know about a particular protocol or check that a specific idea is valid—in the grand scheme of science, only the conclusions (and new techniques) matter. However, we are still stuck in a rather primitive cycle of primary article -¿ follow-up papers -¿ review article -¿ new primary articles (perhaps with a subtle shift in direction prompted by a review/synthesis), rinse and repeat.

Jan Grundemann and i were talking over lunch about the possibility of taking greater advantage of the technological advances in web design, development, and tracking. I use git to help keep track of changes in different coding projects and it seems that science would benefit from a way to both amend articles with new data in a more meaningful way than the current method, which normally results in erratum or corrigendum. For example, if you get new data that enhances a particular hypothesis, this should be amended. A best solution would be to have each article contain a list of bullet points of hypothesis tested and conclusions made. Then these could be easily cross-referenced by other papers specifically, we need to move beyond the horribly vague and sometimes inaccurate tendency to cite a paper without reference to specific passages, conclusions or figures. However, this needs to be done in a way that doesn’t clutter up the page, hence just enhance current citation methods by giving each hypothesis and conclusion in a paper specific labels and other authors can make pointers to them. Or as people who use the internet like to call them, links.

Now back to the current problem of reading papers. The problem is thus: you read a paper hoping to gain another bit of knowledge that can be added to a general framework for understanding a specific problem. This could be the involvement of huntingtin in clathrin-mediated endocytosis or how artificially altering the temperature of the hypothalamus can lead to changes in body temperature. However, while humans are great at forming associations and keeping a rough idea of a theory in mind, they often fail at the specifics without investing endless hours memorizing the content. Considering we are good at creative thinking, linking disparate ideas, and a host of other higher-level processing, the fact that we still spend so much time trying to assimilate information and distill it into a (often too) simple theory is quite astonishing. There is a rather simple solution to the problem.

Wikipedia is a prime example of how the crowd can organize the world’s information. In addition, there are various websites, from NCBI’s to EMBL’s to SGD, that attempt to aggregate the vast amount of biological data available. However, they often lack the key human readable, interpretable, and manipulatable elements needed for the average researcher to both contribute to and benefit from. Thus, a resource needs to be developed that allows each field—be it ventral tegmental area (VTA) contribution to learning or the role of biodiversity in ecosystem stability—to develop a graphic describing the process and allow leaders in the field to edit it after publishing a paper. This would allow a history of changes and links to specific articles supporting or refuting those changes to be easily visible in a way that can lower the barrier to entry—a curated literature review would be available. The only issue is how to segment areas and allow cross-pollination, e.g. if you are studying Huntington at the biochemical level, how do we abstract or allow APIs that interface with those working at the cellular or systems neuroscience levels? I’ll leave that for another post.

go to post.

2013.08.21 biafra ahanonu 2013-08-21T00:00:00+02:00 2013-08-21T00:00:00+02:00

2013.08.21 [link]

useful links

I’ll add these to the useful links main section (see date: ), but they are presented below for quick reference. Thanks to Nobie Redmon for some of these.

go to post.

2013.08.19 biafra ahanonu 2013-08-19T00:00:00+02:00 2013-08-19T00:00:00+02:00

2013.08.19 [link]

brain machine interfaces


Figure 77: Human control of rodents.

Pacific Rim (do yourself a favor and see it, it’s awesome) inspired me to re-look into the state of brain-machine interfaces. The guardian was thinking along the same lines and has a great article on the subject: Are two heads better than one? The psychology of Pacific Rim.

Initial BMI papers showed primitive control of robotic arms.[? ? ] More recently, advances have allowed not only control, but the ability to receive tactile and other feedback.[? ? ? ? ] Surgeon’s have begun to use robotics to perform ever more delicate surgeries, e.g. see the da Vinci surgical system, but the possibility of more precise control via an integration of BMI and hand control might push the technology even further.

A spat of recent papers show that it may be possible to have several brains work together to achieve a common task.[? ? ] Yoo, et al. have also shown that a human can achieve basic control of another animal, in this case a rat, through BMI.[? ] These developments are a nice addition to the plethora of BMI technologies aimed at helping the disabled. While those are worthwhile, expanding BMI to help humans achieve tasks we would otherwise be unable to accomplish could lead to technologies both dreamt of in science-fiction (e.g. Pacific Rim’s jaegars needing two minds to control them) and those that are still unknown-unknowns. For example, imagine if the pilot and co-pilot on an airplane were linked, such that communication errors (see Air France Flight 447) might be avoided.

go to post.

2013.08.18 biafra ahanonu 2013-08-18T00:00:00+02:00 2013-08-18T00:00:00+02:00

2013.08.18 [link]

recording/decoding neuronal populations

An ultra-lightweight design for imperceptible plastic electronics

Steady or changing? Long-term monitoring of neuronal population activity

Neural Syntax: Cell Assemblies, Synapsembles, and Readers

A Family of Algorithms for Computing Consensus about Node State from Network Data

Gain Control Network Conditions in Early Sensory Coding

bonus papers

In vivo synaptic recovery following optogenetic hyperstimulation


We talked a little about terahertz applications in neuro during cs379c. A recent paper in Science goes over terahertz applications for graphene.

Graphene for Terahertz Applications

go to post.

2013.08.13 biafra ahanonu 2013-08-13T00:00:00+02:00 2013-08-13T00:00:00+02:00

2013.08.13 [link]


Graphene for Terahertz Applications

Global Epigenomic Reconfiguration During Mammalian Brain Development

Exotic optics: Metamaterial world

In vivo robotics: the automation of neuroscience and other intact-system biological fields - Boyden discusses advances in automation.

Investigating the role of firing-rate normalization and dimensionality reduction in brain-machine interface robustness - looking at BMI robustness

Relationship between intracortical electrode design and chronic recording function - electrode performance degradation over time is a real issue in BMI and other aspects of neural recording. This paper examines the design of electrodes to improve chronic recording.

On brain activity mapping: insights and lessons from Brain Decoding Project to map memory patterns in the hippocampusOn brain activity mapping: insights and lessons from Brain Decoding Project to map memory patterns in the hippocampus

Connectomic reconstruction of the inner plexiform layer in the mouse retina - Seung and crew continue the grand connectome quest by reconstructing a piece of the mouse retina.

In vivo time-gated fluorescence imaging with biodegradable luminescent porous silicon nanoparticles

Orbitofrontal and striatal circuits dynamically encode the shift between goal-directed and habitual actions - Gremel and Costa develop a novel paradigm to analyze the roles of dorsal medial striatum (DMS), dorso-lateral (DLS) and orbitofrontal cortex (OFC) in habitual and goal-directed activity.

Dopamine Modulates Risk-Taking as a Function of Baseline Sensation-Seeking Trait

Multi-task connectivity reveals flexible hubs for adaptive task control

Micropatterned substrates coated with neuronal adhesion molecules for high-content study of synapse formation

go to post.

2013.08.12 biafra ahanonu 2013-08-12T00:00:00+02:00 2013-08-12T00:00:00+02:00

2013.08.12 [link]

various papers and links

Going to start posting links to papers that might be of interest from a technology or scientific perspective.

Brain Activity in Valuation Regions while Thinking about the Future Predicts Individual Discount Rates

Motor Cortex Feedback Influences Sensory Processing by Modulating Network State

Cellular and Synaptic Architecture of Multisensory Integration in the Mouse Neocortex

The Need for Research Maps to Navigate Published Work and Inform Experiment Planning - nice note about experimental planning and design.

Information and Efficiency in the Nervous System—A Synthesis

IBM Scientists Show Blueprints for Brain-like Computing

Brute force searching, the typical set and Guesswork - maybe slightly off-topic, a layman’s explanation can be found at: Encryption is less secure than we thought

Prolonged dopamine signalling in striatum signals proximity and value of distant rewards

Inorganic materials: Intuition weaved into computation

Network link prediction by global silencing of indirect correlations

Network deconvolution as a general method to distinguish direct dependencies in networks

CRISPR papers Heritable gene targeting in the mouse and rat using a CRISPR-Cas system
Simultaneous generation and germline transmission of multiple gene mutations in rat using CRISPR-Cas systems
Targeted genome modification of crop plants using a CRISPR-Cas system

go to post.

2013.08.05 biafra ahanonu 2013-08-05T00:00:00+02:00 2013-08-05T00:00:00+02:00

2013.08.05 [link]

keeping up with the literature

There is a useful blog called Mo Papers Mo Problems that contains a weekly listing of papers and some insight into recent work.

go to post.

2013.07.31 biafra ahanonu 2013-07-31T00:00:00+02:00 2013-07-31T00:00:00+02:00

2013.07.31 [link]

wireless optogenetics

Zhang’s group recently released optogenetic activation of genes and now there is a paper demonstrating wireless optogenetic control of mice.[? ] Obvious synergies abound, especially when combined with freely moving imaging of mice neurons.

go to post.

2013.07.28 biafra ahanonu 2013-07-28T00:00:00+02:00 2013-07-28T00:00:00+02:00

2013.07.28 [link]

arvix paper

Our paper reviewing scalable technologies in neuroscience is now on arvix.[? ]

go to post.

2013.07.27 biafra ahanonu 2013-07-27T00:00:00+02:00 2013-07-27T00:00:00+02:00

2013.07.27 [link]

basic BRAIN research

Nature neuroscience has a good editorial about basic research (link).

go to post.

2013.07.26 biafra ahanonu 2013-07-26T00:00:00+02:00 2013-07-26T00:00:00+02:00

2013.07.26 [link]


Figure 78: Inducing false memories.

false memories

Ramirez, et al. successfully create a false memory by manipulating hippocampal circuits (link).[? ] The used a fear-conditioning paradigm along with mice expressing c-fos-tTA;TRE on a doxycycline (Dox) diet—an activity dependent promoter that can be suppressed with Dox present, but with Dox absent causes a gene downstream of a tTA responsive element (TRE) to be activated. Injecting AAV-TRE-ChR2-mCherry and removing animals from a Dox diet (these two allow only active neurons to express ChR2-mCherry, which can later be re-activated by light), they associated foot shocks with context B then expose the animal the next day to novel context A. Then showed that they could cause animals to have increased freezing by re-activating cells that were active during fear encoding in context B, even though the animal is in a new context A where no fear training has occurred. The same group conducted a similar study published last year.[? ]

Besides the super cool result, the system developed here could potentially be exploited in other contexts. Perhaps we want to see if a particular structure is involved in reward prediction, in this case the striatum. It might be possible to teach an animal to associate a cue with a reward and label only neurons that respond. You can then re-activate those neurons to see if the animal exhibits stereotyped behavior.

go to post.

2013.07.25 biafra ahanonu 2013-07-25T00:00:00+02:00 2013-07-25T00:00:00+02:00

2013.07.25 [link]

Phonon lasers

Cheng-Hsun Wu showed me the SASER he is building along with Barney Cruz. Got me interesting in looking into phonons and acoustic imaging.

statistics of molecular ticket tape

Glaser, et al. have published[? ] their mathematical modeling of the molecular ticker tape idea that Kording proposed[? ], this helps show the limits of current polymerases and highlights key trade-offs that will need to be taken into consideration. As i wrote in the Google technical report[? ], this technology seems likely to be implemented in the near future and mining of available polymerases or conducting directed evolution experiments to make new ones shouldn’t be too far off.

functional tracing of neural projections

Kawashima, et al. have added another tool to neuroscientists belt in the form of E-SNARE, an activity dependent promoter.[? ] This could potentially be used in combination with a psuedorabies virus[? ] to allow analysis of functional anatomy, that is, the structure of circuits activated during a particular stimulus presentation.

connectivity review article

Mapping the connectivity of neural systems is a priority as it can given an idea of the functional output of a circuit as well as rule out particular hypothesis. Yook, et al. give a nice update on techniques currently employed in connectome reconstruction: EM, Brainbow, array tomography, mGRASP[? ], trans-synaptic tracing, and microscopy-based methods (e.g. SPIM).[? ] For example, CLARITY can be used to clear the brain and SPIM used to map the connectivity is a manner orders of magnitude more rapid than EM.[? ? ? ]

go to post.

2013.07.23 biafra ahanonu 2013-07-23T00:00:00+02:00 2013-07-23T00:00:00+02:00

go to post.

2013.07.22 biafra ahanonu 2013-07-22T00:00:00+02:00 2013-07-22T00:00:00+02:00

go to post.

2013.07.19 biafra ahanonu 2013-07-19T00:00:00+02:00 2013-07-19T00:00:00+02:00

2013.07.19 [link]

nature BRAIN opinion piece

Nature published a recent editorial by Alison Abbott (Neuroscience: Solving the brain) giving a high-level overview of the challenges facing the BRAIN initiative project and some of the available technologies to meet them. It is a good layman’s overview of the technologies, though it fails to mention microscopy[? ] and genetically-modified calcium (or other) indicators (e.g. GCaMP) as a method of measuring neural activity, a glaring oversight.

go to post.

2013.07.18 biafra ahanonu 2013-07-18T00:00:00+02:00 2013-07-18T00:00:00+02:00

go to post.

2013.07.12 biafra ahanonu 2013-07-12T00:00:00+02:00 2013-07-12T00:00:00+02:00

go to post.

2013.07.11 biafra ahanonu 2013-07-11T00:00:00+02:00 2013-07-11T00:00:00+02:00

go to post.

2013.07.08 biafra ahanonu 2013-07-08T00:00:00+02:00 2013-07-08T00:00:00+02:00

go to post.

2013.07.04 biafra ahanonu 2013-07-04T00:00:00+02:00 2013-07-04T00:00:00+02:00

2013.07.04 [link]

sources of information

Update: this has been moved to the beginning of the document and will be regularly updated there.

Finding information on the cutting edge often requires one to go beyond the New York Times and other news sources. For this i turn to a variety of journals, most of which allow me to browse for articles that seem interesting or applicable to neuroscience. While summaries by science journalists are often useful for getting a bit of history and other layman interpretations, it is normally easier to delve right into the publications, reviews, and articles then synthesize what you’ve read and decide whether it is viable based on talking to experts in the field or working through the math, experimental design, etc. Below is a short list of some sources that are useful for staying up-to-date:

go to post.

2013.07.02 biafra ahanonu 2013-07-02T00:00:00+02:00 2013-07-02T00:00:00+02:00

go to post.

2013.07.01 biafra ahanonu 2013-07-01T00:00:00+02:00 2013-07-01T00:00:00+02:00

go to post.

2013.06.30 biafra ahanonu 2013-06-30T00:00:00+02:00 2013-06-30T00:00:00+02:00

go to post.

2013.06.29 biafra ahanonu 2013-06-29T00:00:00+02:00 2013-06-29T00:00:00+02:00

go to post.

2013.06.26 biafra ahanonu 2013-06-26T00:00:00+02:00 2013-06-26T00:00:00+02:00

go to post.

2013.06.20 biafra ahanonu 2013-06-20T00:00:00+02:00 2013-06-20T00:00:00+02:00

go to post.

2013.06.14 biafra ahanonu 2013-06-14T00:00:00+02:00 2013-06-14T00:00:00+02:00

go to post.

2013.06.09 biafra ahanonu 2013-06-09T00:00:00+02:00 2013-06-09T00:00:00+02:00

go to post.

2013.06.06 biafra ahanonu 2013-06-06T00:00:00+02:00 2013-06-06T00:00:00+02:00

go to post.

2013.06.04 biafra ahanonu 2013-06-04T00:00:00+02:00 2013-06-04T00:00:00+02:00

go to post.

2013.05.30 biafra ahanonu 2013-05-30T00:00:00+02:00 2013-05-30T00:00:00+02:00

go to post.

2013.05.29 biafra ahanonu 2013-05-29T00:00:00+02:00 2013-05-29T00:00:00+02:00

go to post.

2013.05.27 biafra ahanonu 2013-05-27T00:00:00+02:00 2013-05-27T00:00:00+02:00

go to post.

2013.05.24 biafra ahanonu 2013-05-24T00:00:00+02:00 2013-05-24T00:00:00+02:00

2013.05.24 [link]

optical readout of brain data


Figure 97: Example OFDM circuit

Was talking in lab to Nobie Redmon about how we would be reading-out 90TB/sec of data if we recorded voltage in every neuron in the human brain (about 1011). We weren’t sure whether it would be best to go an optical route, wirelessly transmit, or some other route. Due to reliability and bandwidth, it seemed that the optical route was the best choice. There are several technologies that can already reach 40GB+/sec, such as PON systems, while experimental tests over several kilometers have shown speeds of 100+TB/s are achievable.[? ] Cvijetic, et al. discuss the advantages of optical orthogonal frequency division multiplexing (OFDM) to increase data rates[? ] and Glick, et al. give a good overview of setting up such a system more broadly.[? ] Many of these technologies take advantage of time-division multiplexing (TDM), which allows for the transmission of multiple signals by dividing indivudal signals in time. For example, if you want to send packet A (8 bit/s) and B (8 bit/s), you could have a TDM device that runs at a much higher rate split the packets in time, e.g. if it could send 16 bit/s, then you could send all of packet A and B in 1 second (A first for 0.5 sec at 16 bit/s followed by B in the same manner), in effect doubling your throughput.

It seems like a feasible goal to miniaturize some of this technology and modify it for use in neurotechnologies. Still reading on the subject and will post more in the future. However, the issue shifts to storage of said data, which is another issue entirely.

analysis of brain data

I have also been wondering about the actual analysis of the data and how that would be dealt with in terms of time. It is well known in electron microscopy that it is not the data collection where time is lost, but the insane amount of time it takes to analyze the data.[? ? ] Researchers like Sebastian Seung and Kevin Briggman are building connectomes while others, like Hanchuan Peng and his SmartScope concept, are working out ways to integrate image acquisition and analysis. This include off-line analysis, such as BrainAligner for Drosophila brains. However, compared to the planned data we could receive from even basic experiments for BRAIN activity mapping, there needs to be order of magnitude leaps in algorithmic speed to deal with it all.

Knowing that Google has to deal with analyzing massive amounts of data on a daily basis—and due in part to methods like MapReduce[? ]—i asked Tom how Google deals with this data deluge. They use a combination of stack optimization to help deal with small files that they need to continuously move around. The small files are normally shards created from larger datasets to help split-up the task of analysis.

Research groups have started to develop architectures to analyze large biology datasets, particularly for next-generation sequencing.[? ] This type of research should be fundamental to any BRAIN initiative project as inevitable people are going to end up with terabytes of data and no way to analyze it in a reasonable time-span or a thought-out manner.


Random eco-evo fact: tardigrades have been found to survive exposure to the vacuum and radiation of space.[? ] It would be super cool to image their neurons after exposure to space, or just culture them to study what properties allow them to survive such harsh conditions. This might lead to insight for both which proteins, pathways, or other biological processes can be up- or down-regulated to cause this insane robustness.

equation test

The below is just a test that the equation conversion works properly, as i’ll be showing my work for back-of-the-envelope calculations in the future.

     1-∑n  ∑K      q       q 2
E  = 2        [yk(x ,w ) - tk]
       q=1 k=1

M  + Qabs = ϵσT 4r + hc(Tr - Ta) + E + C

go to post.

2013.05.23 biafra ahanonu 2013-05-23T00:00:00+02:00 2013-05-23T00:00:00+02:00

2013.05.23 [link]

brain projects across the pond

Interesting talk by Henry Markham (of Blue Brain project fame): A brain in a supercomputer. The European version of BRAIN is Human Brain Project that seeks to develop an exact model of the human brain. This seems to be a bit premature given the current state of knowledge and is unlikely, at that level of detail, to give us any real understanding at the moment. The USA’s focus on developing core technologies—and the light it shines on what we do/don’t know and the analysis of the data that needs to be performed—is a more useful endeavor and will likely yield more long-term insights.5

With regards to a technological singularity, Tom mentioned to read Accelerando. As always, science fiction and futurist predictions are less a desire to accurately predict the future down to the last bolt as it is a way to envision what could happen and how we would deal with this. This is why i love writing short stories.

simulating neural systems


Figure 98: Modeling leech movement circa 70s

A discussion yesterday about simulation of a neural system based on physiological and activity data reminded me of an excellent Science paper from 1978 (yes, ancient in internet-time) called Neuronal generation of the leech swimming movement.[? ] Read this paper for 9.29j at MIT (taught by Michael Fee, who does some amazing work with birds) and we want to see how modeling simple circuits could lead to insight about organization of a system. While the BRAIN initiative hopes to map out many more neurons, old papers like this are instructive in guiding why we need to measure more neurons for longer periods of time and what we will gain from this.

go to post.

2013.05.22 biafra ahanonu 2013-05-22T00:00:00+02:00 2013-05-22T00:00:00+02:00

2013.05.22 [link]

mining the visual system for algorithms

In cs379c we talked with David Cox over at Harvard (check out some of his awesome work!). He works with the rodent visual system in the hopes of mining it for better computer vision algorithms and has developed some pretty sweet methods. Of the questions asked, the one pertaining to how the BRAIN initiative would help him with his algorithms proved quite informative. He mentioned that the focus on more neurons should be complimented by a focus on imaging the same neuron for longer periods of time. I asked briefly about what information he would gain, or how he would adapt his models, should be gain access to this data—the implication was that if you could simulate the data you would get, how would this guide the technology you needed to build. He noted that this might not be the best use of time, but upon prompting him about the development of the nervous system, such as the formation of ocular dominance columns, he stated that it might help inform whether to look at the development of individual elements within a model as it was trained to recognize particular objects.

David also mentioned echo state networks. Not super familiar with them, but will have to look into it more! There is also the possibility of using marmoset's as model organisms, which might help bridge the current gap between primates and lower mammals as funding for chimp and other primate research declines. And the recent hype over CLARITY[? ] raises the question of whether traditional brain slicing (embodied by such companies as NeuroScience Associates and FD NeuroTechnologies) will decline. Given the apparent difficulty of CLARITY, for simple histological checks there shouldn’t be a worry in the medium-term.


Figure 99: Example of the CLARITY technique in action, it allows viewing of stains in the brain without the need to slice it into many sections, potentially allowing for better localization of proteins, etc.

analyzing brain data

An interesting thought game that doesn’t seem to have received rigorous analysis in the whole BRAIN debate is what you would actually do with the increase spatiotemporally and with regards to the number of neurons. Besides the intuitive notion that more is better (America!) there should be a rigorous testing though data scaling or simulating of how analysis would change given a particular increase in the data dimension. For example, what if we could get the response of a million neurons over a 1 second interval?

A side note on data, at a 1 msec interval only looking at spikes (bit = one) or no spikes (bit = zero) would entail 1 bit/msec * 1000 msec/sec * 106 neurons = 109 bits/sec = 120 megabytes/sec, not bad until you want to scale to the human brain (˜1011 neurons), at which point it becomes 11 terabytes/sec, which is an absolutely crazy amount of data for one second—how would you even get it out? Now consider encoding each msec in 8 bits (or 1 byte) to measure membrane potential changes (this would allow for 256 degree change, giving you just enough range over the normal -90mV to ˜160mV of a normal neuron) and you suddenly have 90 terabytes/sec (!). Don’t even ask how we’ll store that, let alone share the data and analyze it (i’ll look into the back-of-the-envelope calculations of basic analysis in a future write-up).

Anyways, back to the main topic. We now have spiking data for a million neurons. What do we do? Put it into some sort of linear classifier, PCA/SVM, factor analysis, k-clustering, or other methods. Undoubtedly with that much data, things are going to group in a higher dimensional space (i’ll actually show this later using randomly generated data in Matlab or R then running each type of analysis on it). But what does that tell us? There are two groups (it appears to me) in the BRAIN discussion: neurobiologist who want to learn something about the mechanism and engineers/computer scientists who want to mine neural systems for new algorithms. Being clear about which goal is being pursued for a particular project will greatly help clarify the types of stimuli, methods of recording, and focus of subsequent analysis. Assuming we want to find out a biological question: we find several levels of sub-circuits that seem to correlate with a behavioral output. Now we need to puterb the system (optogenetics, acoustically, etc.) and observe the changes, ideally in the exact same neurons. This is a monumental technical challenge and what we would actually gain in terms of understanding is not quite clear—at least, for the moment. I’ll have to let this gestate and come back to the topic.

neurons and glia

A side thought i had while talking with Tom was how you would reconcile theories about whether neuronal, glial, or some other form of computation gives rise to particular behaviors? The basic experiment would be to measure neuronal spiking activity or voltage change while simultaneously measuring the membrane potential of electrically coupled glial cells to see which best models the behavior. If the activity of each can equally capture the activity of each, what do you interpret from this? How would you reconcile this redundancy?

power analysis in the brain

The last idea that came to mind was whether we had an accurate method of analyzing how much data we would need to gain a meaningful understanding of the underlying neural system being explored. A key concern of mine was whether we would spend time attempting to make optical, acoustic, magnetic, or eletrical systems to measure neural activity on the 1 msec scale (same scale as an action potential) when what we really need is to only measure at 20 msec resolution (response time of current GCaMP optical activity probes). Using either known simple model organisms or just simulating a basic circuit (there are various models for this that i can get into later) should allow us to progressively lower the temporal resolution and see if we can still capture the system response accurately. Of course, there are great benefits to recording individual action potentials, but as a rigorous test of what technology should be focused on, and thus taxpayers dollars spent on, this analysis should be done.

go to post.

2013.05.20 biafra ahanonu 2013-05-20T00:00:00+02:00 2013-05-20T00:00:00+02:00

2013.05.20 [link]

analyzing OPIDs for brain readout

Yael Maguire came by class a couple weeks ago to chat about using RFID or other wireless technologies (e.g. optical RFID) to measure brain activity. The chief reason for this is that using electrode or other physical channels to extract information out of the brain would not scale properly if you want to measure hundreds, thousands, or millions of neurons. Many of the ideas were pretty neat, but i wanted to do a quick calculation about whether you could feasibly fit nano-OPIDs into the brain, assuming the other technical hurdles were worked out.

The basic schematic for the chip (in a best case scenario) was about 10x10x5 m = 5e-16 m3. Assuming that the human brain contains about 1011 neurons, we have 5e-16 m3* 1011neurons = 5e-05 m3 in total volume for our chips if we want to record from every neuron in the brain. Human brain volume is ˜1450 cm3 = 0.00145 m3. So if we want to calculate the amount of space that our chips would take up: 5e-05 m3/0.00145 m3 = ˜3.4% of total brain volume.

That might not seem like a lot, but for a system as delicately balanced as the brain, that could cause serious problems, the least of which are experimental artifacts. I’ll try to contact a neurologist or search through the literature on brain tumor sizes that cause serious problems to see how this distributed increase in brain volume would disrupt behavior.

go to post.