brain initiative notes

With the increased use of calcium (and voltage) imaging for a plethora of research purposes, the need for tools that can accurately, quickly, and efficiently process this data is needed. As a graduate student in Mark Schnitzer's lab, I built and furthered new tools (e.g. CELLMax and CLEAN) and software packages (see CIAtah, https://github.com/bahanonu/ciatah) to improve analysis of calcium imaging data. While there are existing reviews covering calcium imaging analysis, I decided to expand upon the table in my dissertation to include more recent and additional older algorithms, tools, and analysis pipelines. Those interested in a brief overview of calcium imaging, see the return of the brain initiative notes.

The table below is primarily focused on tools for calcium imaging, but some of them also include tools for voltage imaging, data handling for biology, and more. The goal will be to expand this to include additional imaging analysis tools going forward and I might make it more dynamic akin to my Nobel Prize Browser.

This is a part of the brain initiative notes and can also be found at neuroscience imaging analysis tools or https://github.com/bahanonu/imaging_tools.

Table updates
Below is a changelog of updates to the table, such as adding new methods.

• 2021.11.14 - Added FIBSI, SpecSeg, FIOLA, and PatchWarp.

• 2021.11.29 - Added MVG-CNN.

• 2023.09.12 - Updated online table to include additional papers (added throughout 2022-2023).

Table note

• I use cell extraction to refer to algorithms that perform cell segmentation and extract neural activity traces.

• Any additional papers or algorithms that should be added or suggested updates to the table, leave a comment below or open an issue on the GitHub page, I want to make sure everyone’s brilliant work is acknowledged!

• In cases where the publication did not explicitly give the algorithm a name, made one based on the underlying method used.

• This table includes algorithms that simultaneously extract cell images/contours and reconstruct cell activity traces along with ones mainly focused on determining one or the other.

• Several calcium imaging related packages have also been included along with algorithms dealing with post-hoc handling of data or cell activity traces.

• CELLMax is our algorithm that we have presented at Society for Neuroscience and COSYNE.

With the increased use of calcium imaging for a plethora of research purposes, the need for tools that can accurately, quickly, and efficiently process this data is needed. As a graduate student in Mark Schnitzer's lab, I built and furthered new tools and software packages to improve analysis of calcium imaging data. While there are existing reviews covering calcium imaging analysis, I decided to expand upon the table in my dissertation to include more recent and additional older algorithms, tools, and analysis pipelines. Those interested in a brief overview of calcium imaging, see the return of the brain initiative notes.

The table below is primarily focused on tools for calcium imaging, but some of them also include tools for voltage imaging, data handling for biology, and more. The goal will be to expand this to include additional imaging analysis tools going forward and I might make it more dynamic akin to my Nobel Prize Browser.

This is a part of the brain initiative notes and can also be found at neuroscience imaging analysis tools or https://github.com/bahanonu/imaging_tools.

Note: I use cell extraction to refer to algorithms that perform cell segmentation and extract neural activity traces. Any additional papers or algorithms that should be added or suggested updates to the table, leave a comment below or open an issue on the GitHub page, I want to make sure everyone’s brilliant work is acknowledged!

Table updates Below is a changelog of updates to the table, such as adding new methods.

• 2021.11.14 - Added FIBSI, SpecSeg, FIOLA, and PatchWarp.

• 2021.11.29 - Added MVG-CNN.

Updated calcium imaging analysis tools table (originally created/updated on 2020.02.07). Full article to follow.

Readers can leave comments and ask questions on the bahanonu.com associated blog post.

Large-scale calcium (Ca²⁺) imaging, using the change in Ca²⁺ concentration inside neurons as a relatively direct measure of neural activity, has become an essential tool for neuroscientists to probe neural ensemble activity and identify coding principles the brain uses to process sensory information, store memories, and produce behaviors (Hamel et al., 2015). Originally, chemical fluorescent dyes such as OGB-1 allowed experimenters to measure Ca²⁺ concentration in neurons and thus neuronal activity (Paredes et al., 2008). These synthetic dyes allowed researchers to begin studying detailed intracellular Ca²⁺ dynamics along with activity of multiple neurons at once. However, the dyes had drawbacks: they did not allow long-term (weeks to months) imaging of neural activity, dye loading could be problematic in certain brain cell types (e.g. those with thick cell walls) or regions (deep brain structures), dyes would target non-specifically to various cellular compartments/organelles and would leak out of cells, and they could not be targeted in a genetically defined manner (though there is limited indication that this might be possible (Tour et al., 2007)). On the other hand, genetically encoded Ca²⁺ indicators (GECIs) have helped expand the types of experiments and questions that can be answered using Ca²⁺ imaging (Chen et al., 2013; Dana et al., 2018; Mank et al., 2008) by addressing many of the issues with the Ca²⁺ dyes: using adeno-associated and other viruses, they can be expressed chronically from weeks to years ((Ziv et al., 2013) and personal experiments/observations); specific promoters upstream of the GECI gene can lead to expression in specific cell types; tagging the GECI with certain amino acid sequences can enable targeting to specific regions of the cell (e.g. the axon terminal (Broussard et al., 2018; Dreosti et al., 2009)); and many other advantages. However, once experimenters had collected their Ca²⁺ imaging movies, they faced a formidable challenge in extracting both cell locations/shapes and their associated fluorescence activity traces from each of these movies.

One method of obtaining cell shapes and activity traces is via manual identification and selection of regions-of-interest (ROI) by humans and then taking the average pixel value within said ROI on each frame to reconstruct the activity trace for that cell (ROI) during the entire movie. The manual identification can take a lot of time, requires humans to browse through the entire movie frame-by-frame, and can lead to false negatives especially for low signal-to-noise ratio cells. The later is especially true for one-photon movies and modern GECIs where the baseline is too low to allow cells to be seen except for when they are active. Beyond missing cells, the ROI approach also tends to introduce crosstalk by contaminating each cell’s activity trace with fluorescent signals from nearby or overlapping cells (while cells themselves do not physically overlap, this occurs due to optical limitations and the fact that many Ca²⁺ imaging datasets are 2D, which leads to perceived spatial overlap when a 3D volume is flattened), background (e.g. neuropil from axons and dendrites), blood vessels, and other sources. Thus, a good Ca²⁺ imaging cell extraction algorithm needs to meet several criteria to allow biologists to obtain high-quality data from their movies:

• (1) be automated—reduces time needed to process data, especially as dataset sizes grow, and removes a degree of human introduced variability from manual ROI selection;

• (2) scale well with dataset sizes—this is not required, but preferred as large-scale Ca²⁺ imaging videos are becoming more widespread;

• (3) maintain high fidelity—finding all Ca²⁺ events associated with a particular cell;

• (4) have low crosstalk—minimize number of Ca²⁺ transients or other fluorescent signals in a cell’s activity trace that are not from that cell.

To solve the above problems and meet those criteria, many groups have developed and published new methods. An early popular method used Principal Component Analysis followed by Independent Component Analysis (PCA-ICA) (Mukamel et al., 2009); this was a great step forward and met several ”good algorithm” criteria. However, it could fail in cases with a lot of correlated background noise or in regions of high cell density and activity. Over the next several years, additional techniques were published to tackle the problem and among these non-negative matrix factorization (NMF) (Pnevmatikakis et al., 2014; Maruyama et al., 2014) emerged as a promising candidate to meet many of the ”good algorithm” criteria. After publication, variants and enhancements of the method (CNMF, sc-CNMF, CNMF-E, etc.) sought to improve the cell activity traces by integrating Ca²⁺ dynamics into the model (constrained NMF—CNMF) or to compensate for issues in the original algorithm (e.g. handling of correlated background activity as seen in miniature microscope movies, see CNMF-E). In addition, there is an emerging focus on online (e.g. close as possible to real-time) processing of Ca²⁺ imaging data for closed-loop experiments and also methods to improve compression of data (Buchanan et al., 2018; Giovannucci et al., 2017).

Beyond those algorithms discussed above, a variety of other techniques have been released that either directly allow experimenters to simultaneously obtain cell images and activity traces or that focus on detecting cells or improving already acquired activity trace accuracy. For example, several publications have initial or limited methods for post-hoc automated removal of false positive signals (e.g. PCA-ICA, CNMF, etc. all output signal sources that upon manual inspection turn out to not be cells or other biologically relevant signals [e.g. dendrites]); look for this to be tackled to a greater degree going forward (I will have more on this in the future).

To help new experimenters gain a clearer picture of the landscape, see below table for a list of many different Ca²⁺ imaging cell extraction and activity trace reconstruction algorithms. This table will be updated to include any missing algorithms and further discussion about Ca²⁺ imaging analysis will take place in future posts.

A new paper from my lab has been published characterizing new voltage sensors.

• High-speed recording of neural spikes in awake mice and flies with a fluorescent voltage sensor

• A faster, brighter picture of brain cells in action - brief news article on the paper.

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.

• 457.02 - The impact of pain on motivation

  • Neil Schwarz talked about his recent science paper, see:

  • Decreased motivation during chronic pain requires long-term depression in the nucleus accumbens

• 457.05 - The anterior cingulate cortex as a substrate for chronic pain-induced depression: molecular, lesional and optogenetic evidences.

  • Observe presynaptic NMDAR independent LTP in ACC.

  • Chronic pain leads to an increase in anxiety, involvement of ACC and amygdala

  • They use a sciatic nerve compression chronic pain model and test responses in a variety of assays

    • Dark-light box test

    • Novelty suppressed feeding

    • Splash test

  • In depressed patients, ACC has decreased gray matter volume.

  • Previous studies suggest that ACC encodes aversive component of chronic pain

  • Posterior intralimbic cortex lesion recovers mechanical allodynia post chronic pain

  • Lesion ACC rescues chronic pain induced anxiodepressive behavior. Optoinhibition also rescues depressive-like behavior.

  • Fluoxetine stops change in depressive-like behavior without changes in mechanical withdrawal thresholds, suggesting a role for 5HT/serotinin.

  • Papers

    • Xu, J Neuro 2008

    • Koga, Neuron 2015

    • Liu Lab

    • Barthas 2015

• 457.06 - Nucleus Accumbens subregions dissociate encoding of values for reward and pain

  • Only a small portion of end organ injuries lead to chronic pain, want to find out what predisposes patients to develop chronic pain.

  • Note: not clear if this is a peripheral or central issue, some argue it is purely peripheral and that by fixing peripheral issues, CNS problems will disappear.

  • Connectivity between mPFC-NAcc is increased in patients who develop chronic back pain.

    • Same with white matter diffusivity

    • See Corticostriatal functional connectivity predicts transition to chronic back pain

    • New data from patients in this study indicate that mPFC-NAcc-amygdala separates chronic and sustained back pain groups.

  • Hippocampus and amygdala volume are lower in persisting back pain patients.

  • About 60% of risk of developing persistent back pain in patients is captured by white matter and connectivity between mPFC-NAcc-amygdala circuit.

  • Resting state activity, by around 29 days post SNI, large reorganization in connectivity in rodent models.

  • NAcc shell D2 amplify pain behavior in rodents.

    • 5 days post SNI, see increase in iSPN activity in slice.

    • Modulating iSPN activity with dreads causes changes in pain thresholds, e.g. increase D2 MSN activity leads to decreased thresholds

    • Implies that D2 MSNs can amplify neuropathic pain.

    • Doesn’t seem that they looked before pain to see if there is a change or used optogenetics.

  • papers        

    • Lee, J Neuro 2015

    • Mutso 2012

• 457.07 - Intracellular pathways in the Nucleus Accumbens modulate the antiallodynic actions of antidepressants.

  • See paper: RGS9-2--controlled adaptations in the striatum determine the onset of action and efficacy of antidepressants in neuropathic pain states

    • Looks at RGS9-2 genes role in TCA (Tricyclic antidepressant) responses during neuropathic pain.

    • RGS9KO mice respond earlier to TCAs in terms of mechanical allodynia and immobility

    • RGS9-2 in NAcc blocks Desipramine’s (TCA) antidepressant effects.

      • Desipramine

    • HDAC5 KO more susceptible to antidepressants.

    • Sequenced NAcc after neuropathic pain with and without RGS9 KO.

    • Genes changed in RGS9 KO are the same ones responding to Desipramine.

    • DMI treatment reverses the effect of nerve injury on gene expression.

  • Notes

    • There was a conversation afterwards regarding the effectiveness of various treatments, notes on the drugs mentioned below, will need to verify the claims.

      • Duloxetine - doesn’t work well for treating lupus induced pain

      • Pregabalin - works but with side effects

      • Cannabinoids - see Cannabinoids in Pain Management: An Update

    • If it is immune system mediated, could see effective - Apkarian

    • Why are none of the patients in Apkarian’s study improving? Are they irresponsive to treatment?

    • Do mood disorders go away fairly quickly after pain is treated?

    • Higher depression between transient vs. sustained patients? Apkarian says that neither group had a significant change in the rate/development of depression pre/post study?

• DP03 - Measurement of phasic dopamine signals in the rat nucleus accumbens core and shell in response to noxious stimuli

  • Measured dopamine concentration after noxious stimuli in rat NAcc core/shell using cyclic voltametry.

  • Core has fast, large [DA] transients after noxious stimuli while Shell has smaller, slower to reach max transients.

• 739.06 - Chronic pain causes dysfunction in reward circuitry

  • Opioid engage mesolimbic DA circuitry to produce analgesia.

  • Morphine analgesia blocked by DA antagonists or lesion of VTA-DA neurons.

  • Chronic pain is associated with reduced DA or disruption of mesolimbic circuitry.

  • Opioids fail to stimulate extracellular DA in chronic pain.

  • Cocaine, but not amphetamine, failed to stimulate DA release in chronic pain animals.

  • Suppression of DA tone in chronic pain, since cocaine blocks reuptake and depends on continuous release, would see a deficit there.

  • GABA_A receptor function is altered during chronic pain and there is a decrease in KCC2 expression while BDNF is upregulated in the VTA.

  • KCC2 function compromised in VTA GABA neurons of chronic pain animals.

  • Use CPP to study reward circuit problems during chronic pain.

  • Blocking KCC2 prevents cocaine CPP

  • Papers

    • Suppression of the morphine-induced rewarding effect in the rat with neuropathic pain: implication of the reduction in mu-opioid receptor functions in the ventral tegmental area

    • Microglia Disrupt Mesolimbic Reward Circuitry in Chronic Pain

    • Is there a common molecular pathway for addiction?

• 739.07 - Distinct subpopulations of dynorphin neurons drive aversion and reward

  • Looking at dynorphin and kappa opioid receptor (KOR) function in striatum.

  • KORs are activated in DR and VTA neurons.

  • KOR activity in dorsal NAcc produces preference but in ventral NAcc produces aversion.

  • Wireless u-iLED device for discrete spatial targeting of both dorsal and ventral NAcc in the same animal.

  • NAcc dynorphin-mediated aversion and preference is KOR dependent.

  • Ventral stimulation increased dopamine levels.

  • KOR function is enhanced during pain.

  • Decrease in PR lever pressing with CFA in mice, can rescue with NorBNI (Norbinaltorphimine, opioid antagonist).

    • Interesting that this works, not clear if they tested in SNI models

  • KOR blockade doesn’t also reverse mechanical sensitivity, suggesting that it is exclusive to the motivational aspect.

  • Stimulation of Dyn-Cre NAcc potentiated aversion after CFA.

  • Papers

    • Distinct Subpopulations of Nucleus Accumbens Dynorphin Neurons Drive Aversion and Reward

    • Inflammatory Pain Promotes Increased Opioid Self-Administration: Role of Dysregulated Ventral Tegmental Area mu Opioid Receptors

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

• M16 - Optogenetic inhibition of specific populations of sensory neurons mediating bladder nociception

  • Using SNS-Ai35 mice, they show that they are able to inhibit bladder responses.

  • These are initial results, seems like it might not be super reliable, but they need higher numbers of mice to tell.

• AA35 - Two-photon imaging of light-induced nociceptive processing In vivo

  • Can image DRG neurons at some time as stimulating SNS-ChR2 in mice

  • GCaMP6m (calcium indicator) signal looks okay, DRGs only appear to respond to 10Hz stimulation.

• 

• 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

    • Llewekyn 2010 - Orderly recruitment of motor units under optical control in vivo

    • http://www.nature.com/nm/journal/v16/n10/full/nm.2228.html

• 549.03 - Optical control of stem cell derived motor neurons restores function to paralysed muscles

  • Trying to see whether stem cell derived MNs can be functional.

  • See that grafted MNs both are able to survive and their axons follow the old path of previous MNs’s axons to innervate muscle and form stable neuromuscular junctions (NMJs).

    • They are using embryonic MNs.

  • Using optogenetics, they are able to show graded recruitment of motor units, something that doesn’t seem possible with electrical stimulation in this situation.

  • Chronic stimulation strengthens ES-MN neuromuscular junctions, almost as if they are being exercised

  • Notes

    • https://en.wikipedia.org/wiki/Acute_coronary_syndrome

    • http://cbl.eng.cam.ac.uk/pub/Public/Wolpert/Publications/HamJonWol04.pdf

• 

• 549.04 - New paradigms in wireless light delivery

  • See recently published work: http://www.nature.com/nmeth/journal/v12/n10/full/nmeth.3536.html

  • 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.

• 

• 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 http://www.nature.com/nrn/journal/v1/n2/fig_tab/nrn1100_101a_F3.html

  • 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.

• 

• 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.

    • see http://www.ncbi.nlm.nih.gov/pubmed/24726382

  • 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 http://elifesciences.org/content/4/e09674

    • 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        

    • Unbiased classification of sensory neuron types by large-scale single-cell RNA sequencing

• 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

    • http://www.sciencemag.org/content/340/6129/211.figures-only

  • RF powered LED device, see Neurolux:

    • (http://www.neurolux.org/)[http://www.neurolux.org/]

  • 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.

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 Häusser) gives a nice overview of several recent technologies in the area and future directions.

• All-Optical Interrogation of Neural Circuits

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.

• Science Isn’t Broken

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

• Lighting The Brain - Karl Deisseroth and the optogenetics breakthrough.

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.

• NIH program on Rigor and Reproducibility

• Drug development: Raise standards for preclinical cancer research - a highly cited paper that sounded an alarm about the reproducibility of pre-clinical research.

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

• Statistics for biologists

update Was talking to several people. Pubmed Commons was suggested as a way to both improve reproducibility and provide an avenue for people to provide updates about a data set as they conduct more experiments. Whether people will have the inclination or time is another matter.

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.

• Working with Big Data in MATLAB

• SDCube and hybrid data storage

• Adaptive informatics for multifactorial and high-content biological data

• Considerations for developing a standard for storing electrophysiology data in HDF5

• Making data sharing work: The FCP/INDI experience

• Big data, but are we ready?

• Big data from small data: data-sharing in the `long tail' of neuroscience

• A Standard for Neuroscience Data

• Electrophysiology Task Force

• BrainFormat

• Data Sharing for Computational Neuroscience

• Trends in the production of scientific data analysis resources

• NIH Working Group on Data and Informatics NIH Big Data to Knowledge (BD2K) initiative

• Global Alliance for Genomics and Health

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.

• Spatiotemporal Control of Opioid Signaling and Behavior - optoMOR (mu opioid receptor)

• Natural Neural Projection Dynamics Underlying Social Behavior (figure 8) - optoD1 (dopamine 1 receptor)

• Temporally precise in vivo control of intracellular signalling - opto-2AR and opto-1AR (adrenergic receptors)

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.

• The Neuroterrain 3D Mouse Brain Atlas

• Dr. Salzmann's work at Max Planck Institute for Biological Cybernetics

• Mouse Imaging Centre

• Allen Institute Brain Explorer

• Digimouse: 3D Mouse Atlas

A group in China has applied CRISPR to a fertilized (albiet sterile since it had two sperm enter) human embryo.

• CRISPR/Cas9-mediated gene editing in human tripronuclear zygotes

There have been numerous calls to stop editing of the human germline.

• Chinese scientists genetically modify human embryos Don’t edit the human germ lineScientists sound alarm over DNA editing of human embryos

There is the fear, probably reasonable, that if applied to humans without proper understanding of the effects of these CRISPR-induced modifications, that those modified people could spread these potentially harmful genes long before we figure out the subtle negative effects. Another, more distant, worry is whether China or another nation will attempt to ‘improve’ people by combining increased knowledge about genetic determinants of particular physiological or psychological traits (e.g. athleticism and intelligence) with the ability to modify those specific genetic elements in embryos. Below are a couple links to articles and institutes related to this topic.

• 'Smart genes' prove elusive First GWAS hits for cognitive ability Social Science Genetic Association Consortium Playing with Fire: Linking Intelligence to Our Genetics BGI Cognitive Genomics

Whether this is a political play or not by New Gingrich, but that scientific funding is receiving attention is good news. Maybe it will lead toward more discussions about why the NIH budget has been stagnant and ways it can be increased or better utilized given current funding constraints throughout the government.

• Newt Gingrich: Double the N.I.H. Budget

• High-resolution whole-brain staining for electron microscopic circuit reconstruction

There has been a recent push to make sure that the statistics reported in papers crosses the Ts and dots the Is. I wrote about this last month, see Significance testing.

There is a constant tension between people’s desire for novel, groundbreaking results and solid, reproducible data. While some claim that reproducibility is an ideal that might not be reached, at the very least there should be a consistency within papers when displaying the stats, specifically there should be a need for people to explain the experiments that were tried leading up to the work done in the paper, so people can get an idea of the true-negative rate in studies.

• Too Much Success for Recent Groundbreaking Epigenetic Experiments

• When Are Results Too Good to Be True?

• Mud Sticks: On the Alleged Falsification of Mendel's Data

• Scientific method: Statistical errors

• The Abuse of Power: The Pervasive Fallacy of Power Calculations for Data Analysis

• Testing a Point Null Hypothesis: The Irreconcilability of P Values and Evidence

• Effect size

• It's the Effect Size, Stupid What effect size is and why it is important

• An exploratory test for an excess of significant findings

This looks like an interesting workshop for those that can attend: Neural circuits and behaviour of Drosophila.

Rather than go into crazy detail here, I’ll just post a link to an interesting article on the Fairhall lab website that has short pieces by various successful female scientists and how they balance family and career.

• How does she do it?

A couple weeks ago Virginia Man-Yee Lee (UPenn) gave a talk mainly focused on tau and α-synucleun propagation in the brain along with the emerging strain hypothesis.

A new paper out from Ron Kopito’s lab (Stanford) shines light on an interesting. A half-decade ago a perspective paper from Ron’s lab laid out a prion hypothesis for the transmission of protein aggregates that help explain some of the neuropathology seen in neurodegenerative disorders (see Prion-like transmission of protein aggregates in neurodegenerative diseases).

• Prion-like transmission of neuronal huntingtin aggregates to phagocytic glia in the Drosophila brain

While an increasing number of schools reward students for competitive fellowships, there is a question of whether . A series of studies in the 70s and 80s pointed toward a counterintuitive notion: that introducing monetary rewards actually decreased performance, and this occurred for creative endeavors as well.(Kohn, 1993a,b)

On the other hand, there are a number of meta-analyses that point toward the opposite conclusion, that rewards if properly done can enhance performance/attitude or have no change. I view no change as a negative, as you are erroneously spending capital that could be used elsewhere, but the fact that some studies found positive correlations is enough to give one pause and reconsider the previous arguments given.

There is a question also of whether only having a small reward is somehow detrimental, in a counterintuitive manner. Studies seem to indicate that this might be the case.(Gneezy and Rustichini, 2000) However, the whole field seems to be littered with subtle effects that might either be due to the artificial environment, the manner in which the incentives were worded, and other experimental artifacts. However, the psychological argument, that people will replace or confuse the primary motivation with the monetary one, is compelling and is enough to make one pause before developing an incentive program for a business, program, or institute.

• Why Incentive Plans Cannot Work

• For Best Results, Forget the Bonus

• Reinforcement, Reward, and Intrinsic Motivation: A Meta-Analysis

• Detrimental effects of reward: Reality or myth?

• Pay Enough or Don't Pay at All

• Cerebral cartography: a vision of its future

• The BRAIN Initiative: developing technology to catalyse neuroscience discovery

• Brain/MINDS: brain-mapping project in Japan

• The future of human cerebral cartography: a novel approach - by Markram, whose helping head Europe’s Human Brain Project.

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.

• Consciousness: here, there and everywhere?

• Why I Am Not An Integrated Information Theorist (or, The Unconscious Expander)

• Consciousness: Confessions of a Romantic Reductionist

• Phi: A Voyage from the Brain to the Soul

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.

Reviews

• Experimental evidence needed to demonstrate inter- and trans-generational effects of ancestral experiences in mammals - an interesting perspective article that helps outline and clarify several aspects of the transgenerational epigenetic debate.

• Transgenerational Epigenetic Inheritance: Myths and Mechanisms - a nice review that both defines some terms and lays out the current evidence for transgenerational inheritance in humans and other organisms.

• Epigenetic mechanisms underlying learning and the inheritance of learned behaviors

• Lamarck revisited: epigenetic inheritance of ancestral odor fear conditioning

• Molecular mechanisms for the inheritance of acquired characteristics—exosomes, microRNA shuttling, fear and stress: Lamarck resurrected?

• Environmentally induced epigenetic transgenerational inheritance of disease susceptibility

• Epigenetic memory: the Lamarckian brain

• A lingering smell? - a nice perspective on the Dias, 2014 paper.

• Understanding transgenerational epigenetic inheritance via the gametes in mammals

• Transgenerational Epigenetic Inheritance: Prevalence, Mechanisms, and Implications for the Study of Heredity and Evolution

• Bridging the transgenerational gap with epigenetic memory

• Epigenetic and epigenomic variation in Arabidopsis thaliana - this review looks at epigenetic changes in Arabidopsis thaliana, a commonly used model organisms in plant research.

• Epialleles in plant evolution

Articles

• Parental olfactory experience influences behavior and neural structure in subsequent generations - perhaps one of the stronger transgenerational epigenetic papers, mostly because they demonstrate an actual function behavioral change in offspring that would potentially be advantageous in the wild. In addition, they demonstrate a morphological change that explains these behavioral changes, namely alterations in the olfactory neuron gene expression and the size of olfactory glomeruli.

• Paternally Induced Transgenerational Environmental Reprogramming of Metabolic Gene Expression in Mammals - demonstrates that altering a parents environment (specifically causing them to be on a low protein diet) can cause alterations in their offspring’s phenotype. These changes in the offspring are correlated with alterations in DNA methylation, suggesting a mechanism for transmission.

• Ancestral dichlorodiphenyltrichloroethane (DDT) exposure promotes epigenetic transgenerational inheritance of obesity - suggest that the pesticide DDT can promote obesity in subsequent generations. While it is claimed that this is due to transgenerational epigenetic effects, it is also possible that germline mutations occurred due to exposure to the drug.

• Environmentally Induced Transgenerational Epigenetic Reprogramming of Primordial Germ Cells and the Subsequent Germ Line - investigates how agricultural fungicide vinclozolin can cause epigenetic changes, by altered DNA methylation patterns and transcriptomes, in subsequent generations.

• Epigenetic Transmission of the Impact of Early Stress Across Generations - show that chronic stress early in life, in mice, can cause DNA methylation changes and altered behavior of offspring that are reared normally.

• Epigenetic inheritance of a cocaine-resistance phenotype - show that ingestion of cocaine by parents can affect offpsring resistance to cocaine and alter prefrontal gene expression patterns.

• Paternal Stress Exposure Alters Sperm MicroRNA Content and Reprograms Offspring HPA Stress Axis Regulation - this study interestingly found that microRNAs in sperm where changed after exposure to stress and that offspring of stressed animals showed hypothalamic–pituitary–adrenal (HPA) axis (involved in stress regulation) dysregulation. It’s possible that changes in microRNA level can alter early gene expression patterns, leading to subtle changes in offspring’s later health.

• Implication of sperm RNAs in transgenerational inheritance of the effects of early trauma in mice - this paper also looks at how early trauma can lead to changes in microRNA expression.

• Mechanisms of epigenetic memory and addiction

While optogenetic activation of neurons and other cell types is currently in vogue, most stimulation requires invasive implantation of light delivering fibers and viral expression of channelrhodopsin or other proteins that can modify activity. These both can potentially alter the brains function and lead to a less robust understanding of how the circuit contribute to internal states and behavior. A holy grails of manipulating, and extracting information from, neuronal circuits would be through fully non-invasive methods.

Transcranial magnetic stimulation allow non-invasive stimulation of the brain, but it is relatively non-specific in terms of stimulating specific neuronal populations, whether they be genetically, anatomically, projection-specific, or another subset. This reduces it usefulness for probing causal roles of specific neuronal subsets altered during diseases, such as striatally projecting cortical neurons in psychiatric disorders.

On the other side of things several technologies currently allow for non-invasive probing of brain activity (note, this is different from neuronal activity in many cases): functional magnetic resonance imaging (fMRI), functional near-infrared spectroscopy (fNIRS), electroencephalography (EEG), and positron emission tomography (PET). However, many of these are limited by either not directly measuring neuronal activity (e.g. fMRI and fNIRS); only measuring bulk, non-localized activity (EEG), or measuring the binding of a specific molecule (PET). Recent papers from the Jasanoff lab and others have demonstrated that MRI contrast agents can be made sensitive to specific neurotransmitters (such as dopamine and serotonin) along with other bioactive molecules. This might someday allow imaging of specific neurotransmitter levels in humans one day, which could potentially allow us to gain a greater understanding of people at risk for particular behaviors, e.g. if they have aberrant dopamine levels in the NAcc or elsewhere, we could keep an eye out for them in case they soon develop gambling or other habits related to altered reward processing.

This area of neuroscience research is quite rich and if we could get more precise magnetic stimulation, we could someday have a much better toolkit to probe and measure neuronal activity.

• Wireless magnetothermal deep brain stimulation

• Treating brain disorders with neuromodulation

• Structure-Guided Directed Evolution of Highly Selective P450-Based Magnetic Resonance Imaging Sensors for Dopamine and Serotonin - one of the original papers characterizing the development of contrast agents to help detect neurotransmitters levels using MRI.

• Molecular-Level Functional Magnetic Resonance Imaging of Dopaminergic Signaling

• Advances in using MRI probes and sensors for in vivo cell tracking as applied to regenerative medicine - discusses some applications of new MRI contrast agents (probes) that will help allow imaging of specific cells in vivo, among other applications.

• Dendrimeric calcium-responsive MRI contrast agents with slow in vivo diffusion - interesting concept to image calcium concentrations using MRI.

• Magneto-fluorescent core-shell supernanoparticles - these can be used to do both optical and magnetic resonance imaging.

• Dysfunction of Cortical Dendritic Integration in Neuropathic Pain Reversed by Serotoninergic Neuromodulation

• Optogenetic Recruitment of Dorsal Raphe Serotonergic Neurons Acutely Decreases Mechanosensory Responsivity in Behaving Mice

• Transmitting Pain and Itch Messages: A Contemporary View of the Spinal Cord Circuits that Generate Gate Control

• Identification of Spinal Circuits Transmitting and Gating Mechanical Pain

• Targeted Ablation, Silencing, and Activation Establish Glycinergic Dorsal Horn Neurons as Key Components of a Spinal Gate for Pain and Itch

• Genetic Control of Nociceptive Sensory Neuron Development and Pain Behavior pm Grantome

• In it to win it: Pathway to scientific independence by the NIH K99 award

• NIH RePORTER

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.

• Explicit memory creation during sleep demonstrates a causal role of place cells in navigation

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.

• Optogenetic stimulation of a hippocampal engram activates fear memory recall.

• Creating a False Memory in the Hippocampus

• Bidirectional switch of the valence associated with a hippocampal contextual memory engram

• Engineering a memory with LTD and LTP

Sometimes it’s interesting to see what the most highly cited research is...

• The Agony and the Ecstasy— The History and Meaning of the Journal Impact Factor

• Does Google Scholar contain all highly cited documents (1950-2013)?

• The top 100 papers

• What Are The Most Cited Research Papers Of All Time?

Kay Tye has a review on amygdala circuits that is worth a read, as it gives a nice overview of some history and current opinions/experiments in the literature. However, I think it also represents a certain local minimum in neuroscience that we will likely someday have to overcome if we are to truly understand how the brain performs computations that lead to behavior: the focus on arrow diagrams, the lack of showing the same behavioral experiments done where another brain region is manipulated instead of the BLA (e.g. with optogenetics),super simplified neuronal circuits (often with two classes of neurons explaining the behavior), and no discussion on quantitative predictive power for any models shown.

• From circuits to behaviour in the amygdala

• From circuits to behavior: a bridge too far?

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.

• Psychology Journal Bans Significance Testing

• Basic and Applied Social Psychology Editorial on the null hypothesis significance testing procedure

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¡0.05 and then proceed to publish. It is also a fact that new, positive finds, whether true or not, are more likely to garner attention and publication than a null result. This is a complex issue that will probably grow to dominate scientific discussion in the coming decade.

• Why Most Published Research Findings Are False

• Power failure: why small sample size undermines the reliability of neuroscience

• Scientific method: Statistical errors

• False-Positive Psychology: Undisclosed Flexibility in Data Collection and Analysis Allows Presenting Anything as Significant

• How Many Scientists Fabricate and Falsify Research? A Systematic Review and Meta-Analysis of Survey Data

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.

• In vivo Calcium Imaging to Illuminate Neurocircuit Activity Dynamics Underlying Naturalistic Behavior

• Tools for Resolving Functional Activity and Connectivity within Intact Neural Circuits

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.

• Swept confocally-aligned planar excitation (SCAPE) microscopy for high-speed volumetric imaging of behaving organisms

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.

• Laminar optical tomography: demonstration of millimeter-scale depth-resolved imaging in turbid media

• Laminar optical tomography: high-resolution 3D functional imaging of superficial tissues - gives a nice overview of the technique in the methods section.

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.

• Coupling Mechanism and Significance of the BOLD Signal: A Status Report

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: Using Large-scale Dimensionality Reduction Algorithms for Localization

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.

• The Fundamental Physical Limits of Computation

• Ultimate physical limits to computation

• Limits on fundamental limits to computation

• Fundamental Limits to Moore's Law

• Thermodynamic Cost of Reversible Computing - see layman’s overview Computers Faster Only for 75 More Years

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

• Transgenic Mice for Intersectional Targeting of Neural Sensors and Effectors with High Specificity and Performance

• JAX Allen Mice - mice available here.

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.

• Restricted and Regulated Overexpression Reveals Calcineurin as a Key Component in the Transition from Short-Term to Long-Term Memory

• Genetic and Pharmacological Evidence for a Novel, Intermediate Phase of Long-Term Potentiation Suppressed by Calcineurin

• Overexpression of calcineurin in mouse causes sudden cardiac death associated with decreased density of K+ channels

• Cardiac function and electrical remodeling of the calcineurin-overexpressed transgenic mouse

• Time-dependent systolic and diastolic function in mice overexpressing calcineurin

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.

• High-resolution reconstruction of the beating zebrafish heart

• Rapid 3D light-sheet microscopy with a tunable lens

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.

• Fast two-layer two-photon imaging of neuronal cell populations using an electrically tunable lens

• Tunable Optics - nice overview of tunable optics.

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.

• Human-level control through deep reinforcement learning

• Playing Atari with Deep Reinforcement Learning - original arXiv paper.

• Artificial intelligence: Learning to see and act - a good layman’s overview.

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.

• The AI Revolution: The Road to Superintelligence

• Google’s Grand Plan to Make Your Brain Irrelevant

• Superintelligence: Paths, Dangers, Strategies - pretty good overview of many concepts of superintelligence, how it could arise and evolve, and potential mechanisms to ensure that it doesn’t go rouge.

• The Coming Technological Singularity: How to Survive in the Post-Human Era

• The Singularity Is Near: When Humans Transcend Biology

• The Quest for Artificial Intelligence

• We Live in a Jungle of Artificial Intelligence that will Spawn Sentience

• The AI-Box Experiment - explores the question of whether an AI can convince a human to let it out of its safety box (e.g. the box used to prevent it from directly interacting with the outside world.)

• The Basic AI Drives - interesting exploration of different drives that AI have.

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 http://en.wikipedia.org/wiki/2010_Flash_Crash. Some articles exploring that event in more detail below.

• What caused the flash crash? One big, bad trade

• Findings Regarding The Market Events Of May 6, 2010

• Flash Crash: Could it happen again?

• The monolith and the markets

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.

• Suprachiasmatic stimulation phase shifts rodent circadian rhythms

• The role of the suprachiasmatic nuclei in the generation of circadian rhythms in the golden hamster, Mesocricetus auratus

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.

• Thirst driving and suppressing signals encoded by distinct neural populations in the brain - the new paper

• Why Do We Feel Thirst? An Interview with Yuki Oka Anteroventral Wall Of The Third Ventricle And Dorsal Lamina Terminalis: Headquarters For Control Of Body Fluid Homeostasis?

• Angiotensin, Thirst, and Sodium Appetite

• Drinking behavior following electrical stimulation of the subfornical organ in the rat. - paper from 1983 where they do a similar study only using eletrical stimulation.

• Metabotropic glutamate receptors in median preoptic neurons modulate neuronal excitability and glutamatergic and GABAergic inputs from the subfornical organ.

• Endogenous angiotensin II facilitates GABAergic neurotransmission afferent to the Na+-responsive neurons of the rat median preoptic nucleus.

• Angiotensinergic and cholinergic receptors of the subfornical organ mediate sodium intake induced by GABAergic activation of the lateral parabrachial nucleus.

• Activation of the renin-angiotensin system, specifically in the subfornical organ is sufficient to induce fluid intake.

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.