Medical What is the IBM/EPFL Blue Brain Project?

[hypnagogue]

IBM and The Ecole Polytechnique Fédérale de Lausanne (EPFL) are today announcing a major joint research initiative – nicknamed the Blue Brain Project – to take brain research to a new level.

Over the next two years scientists from both organizations will work together using the huge computational capacity of IBM’s eServer Blue Gene supercomputer to create a detailed model of the circuitry in the neocortex – the largest and most complex part of the human brain. By expanding the project to model other areas of the brain, scientists hope to eventually build an accurate, computer-based model of the entire brain.

http://domino.research.ibm.com/comm/pr.nsf/pages/rsc.bluegene_cognitive.html

see also http://bluebrainproject.epfl.ch/

 

[DocToxyn]

Thanks for the update and the links. Those are some beautiful pictures of neurons in the first link. It will be interestering to see what comes out of this project.

 

[saltydog]

http://domino.research.ibm.com/comm/pr.nsf/pages/rsc.bluegene_cognitive.html

see also http://bluebrainproject.epfl.ch/

This was a quote in the first reference:

"The first phase of the project will be to make a software replica of a column of the neocortex."

"Software replica". That's exciting. Might you have additional information on how that is being accomplished?

Edit: I found this link:
The Blue Column

 

[Moonbear]

Those look like some sort of golgi stain, don't they? Is there a method of doing fluorescent golgi staining? When I look at the "magnified image," the one with orange cells seems to be a confocal z-series overlaid to reconstruct an image through the full depth of the section. Are there other dyes that are readily taken up in post-mortem tissue? I know many ways you can get images like that from animal tissues, but I'm trying to figure out what they could have used as a post-mortem method of staining cells that well. It seems like a formidable task to try mapping the human brain first. It seems we know an awful lot more detail about the circuitry of rodent brains. We're only starting now to attempt to confirm that cell groups present in hypothalamic nuclei of our animal models are also present in human brain, and can't get tissue of sufficient quality to even examine close contacts of fibers with cell bodies let alone confirm synaptic contacts or trace those fibers to the cells of origin.

Well, it does have a cool gee whiz factor to it at least.

 

[neurocomp2003]

the topic was in one of those science mags in may or june.

 

[DocToxyn]

Those look like some sort of golgi stain, don't they? Is there a method of doing fluorescent golgi staining?

Perhaps it was originally a true Golgi stain that was then digitized and colorized to make it cool and pretty. Too bad they didn't have any Purkinje neurons in those images, they have the most gorgeous arborization; everybody loves the cortex, what about other regions?.

 

[Moonbear]

Perhaps it was originally a true Golgi stain that was then digitized and colorized to make it cool and pretty. Too bad they didn't have any Purkinje neurons in those images, they have the most gorgeous arborization; everybody loves the cortex, what about other regions?.

Other regions? For a computer project? They're not as simple, organized, or repetitive. But, I agree, I'd much rather have a computer program that maps the hypothalamus (my own personal preference). But, I guess those of us sitting around cutting brains for a living need to give them the data to input first.

 

[neurocomp2003]

hypothalamus? nah i'd to the imagery center P-O Lobe regions (7a,LIP) or the PVC(wait is that the one they are modelling) and PAC.

 

[Moonbear]

hypothalamus? nah i'd to the imagery center P-O Lobe regions (7a,LIP) or the PVC(wait is that the one they are modelling) and PAC.

If you're going to model cortex, at least do the PFC; that's where all the fun is.

 

[neurocomp2003]

yeah but how do you model the PFC if that's the last thing a child develops i guess the PFC and 7a/LIP go hand in hand but if i remember correctly 7a/LIP develops first.

anyhow they should start with THalamus...good old relaystation. Thats where I'm starting my model...then i move onto the PVC and hopefully PAC(if i can learn sound programming)...don't know how to model PSC because i don't hav ea robot to work with but meh...i'ma just start with vision and audition and move onto imagery

 

[neurocomp2003]

oh and maybe you can help moonbear...i have a brain map that i made from Photoshop(took em too long but because i don't have other software like XML)

these are the brain parts i have so far. I need help with the cortices area because that's where all my textbooks get confusing

SENSES:
eyes/ears/sensorimotor/speech

HindBrain
SupOlive, CochNucleus,TrigeminalNuc, RAF,Cerebellum/pons/medulla.

MidBrain
Colliculus inf/sup. Periaqueductal gray/RedNucleus/Substantia Nigra

Thalamus
LGN/LPN/PUlv/MGN/VPN/ILN-PFN/ANT/VLN/VAN/MDN
-i'm missing hypothalamus/pit.gland but i can't find where they input/output to

Basal Ganglia
-Striatum(Caudate/Putamen), G.P

Limbic System
-Cingulate Cortex(AntCC,PostCC/Retrospinial) ParaHC, RhinalCortex(Peri,Ento)
-Hippocampus(DG,CA3,CA1) Subiculum Layer(SB,PreSB,ParaSB)
-Amygdala,Fornix,septum,MamBodies,

Cortices(where i get lost-primary/secondary/association
- PVC(V1)/SVC(V2,v3,v4,v5),IT,TE,TEO,MT,MST
- PAC(A1)/SAC(A2), ?
- PSC,SSC,?
- PMC/SMC(SMA,PreMC,Cingulate MotorCortices)
- Parietal/Occipital regions 7A/LIP

Some area that i read about that i don't know where to include
Lingual Gyrus,Fusiform,angular,Lemniscus

if you know where these input/output to and can tell me i'd be much obliged

and if you see things that i may have duplicated(by names) or things I need to add please do.
best

 

[Moonbear]

if you know where these input/output to and can tell me i'd be much obliged

Um, that's what DocToxyn and I are talking about. It's REALLY more complex than that. There's no single input/output, and there are a LOT more nuclei than on your list. For example, for the limbic system, you have listed the cingulate, but didn't list the infralimbic and prelimbic cortex, which are also located in the prefrontal cortex.

Even though it's not the human brain, I suggest you get the most recent rat brain atlas: The Rat Brain in Stereotaxic Coordinates, 5th Edition from Paxinos and Watson from Elsevier Publishers, 2005. It comes complete with a CD Rom version of all the diagrams (they're all in coronal view). Plus, this new version was printed in a more compact size than previous versions that don't leave you with any desk space left when it's open. There are human brain atlases too, but they are not as detailed, and I work back and forth between the two when working in human tissue.

Oh, and don't forget the CSF! That's what I'm currently working on, neurons that communicate via CSF. It's not just synapses anymore!

Do you attend the Society for Neuroscience meetings? This year's meeting is in Washington, DC and the registration opens in a week or two. I haven't paid attention to whether there is much in neurocomputing there (I don't think there is), but there are a lot of neuroanatomists there.

I take it you're modeling neural development if you're saying you can't model the hypothalamus because it develops last?

 

[neurocomp2003]

thanks for you reply moonbear...
i'm trying to do a software model from a child development standpoint rather than a hardware model...that is to say I only care about the connections made(graph theory) rather than the the system in which it exists in(the brain structure). So i guess it'd be consider more neurocognitive modelling than brain modelling.

My original undergrad thesis had me starting at the Hippocampus/Subiculum/PFC for spatial navigation. However i wasn't to familiar with C/C++/3D graphics programming at the time so i dropped it in my finally year after 2 years of preparatory research in the topic...didn't realize i should have focused on teh programming rather then the math/psych/physics. Now I am and so i want to start my code where teh child begins to interact with the world, the senses(i can't really do proprioceptive or touch because i don't have a robot to work with) so i want to code the early vision/early audition connections before I tackle the more complex stuff like spatial navigation/imagery/memory/language/consciousness. I'm going to ignore emotions for now.


"Oh, and don't forget the CSF! That's what I'm currently working on, neurons that communicate via CSF. It's not just synapses anymore!" can you reference me to some papers?

best
jack

 

[DocToxyn]

Pretty ambitious work neurocomp, especially since you are looking at development - any specific window? Don't forget the VTA, nucleus accumbens, area postrema...man it goes on forever. Another source for some interesting rodent brain info is the Mouse Brain Library (have you seen this Moonbear?). I sometimes use it when I don't have an atlas handy, and its updated regularly, check it out.

 

[saltydog]

Oh, and don't forget the CSF! That's what I'm currently working on, neurons that communicate via CSF. It's not just synapses anymore!

Would you kindly explain what CSF is?
A reference too wouldn't hurt.

 

[Moonbear]

Another source for some interesting rodent brain info is the Mouse Brain Library (have you seen this Moonbear?).

No, I wasn't aware of that. I've never been far from an atlas, so never considered looking online for a mouse brain atlas. I know Larry Swanson has been pushing for sometime now to get some sort of data base set up and available along the lines of what GenBank is for genes but for the brain. But, he doesn't just want a crude atlas, but something that's searchable by cellular phenotype. I think he just wanted to start with rat brain because that's where we have the most data. So, the idea would be you could type in, for example, arcuate nucleus, and get a list of cellular phenotypes present there, or you could then search a specific phenotype, say dynorphin neurons in the arcuate, and get a list of all the cells and regions that are known to be efferent and afferent to that dynorphin population.

Would you kindly explain what CSF is?
A reference too wouldn't hurt.

CSF is cerebrospinal fluid. There has been a slowly[/size] growing body of literature indicating that CSF may not just be providing a nutrient support for neurons, but actually carry neurotransmitter signals throughout the brain. So far, the data in support of this includes detection of neurotransmitters and neurohormones in CSF of the cerebral ventricles and anatomical evidence of neurons with axon terminals projecting to the cerebral ventricles (anatomically positioned so they could release directly to CSF) and dendrites projecting out into the periventricular regions (anatomically positioned so they can detect signals from CSF). Direct evidence that this is a functional method of communication between neurons within the brain is far more limited. The best direct evidence is work done with suprachiasmatic nucleus (SCN) transplants. This is a small nucleus located just ventral to the third ventricle that function in circadian timing. You can dissect it out of an animal with a mutation that shortens their circadian rhythm to 20 hours rather than 24 hours, encase it in a membrane that allows diffusion of proteins, but does not permit axon re-growth past the membrane, transplant it into another animal (wild type rhythm of 24 hours) with the SCN lesion, and in the recipient, you get restoration of circadian rhythms, but with the short cycle of the donor animal. Silver R, LeSauter J, Tresco PA, Lehman MN. 1996 A diffusible coupling signal from the transplanted suprachiasmatic nucleus controlling circadian locomotor rhythms. Nature 382:810-3.

And then a review article on the system I'm currently studying:
Skinner DC, Caraty A. 2002 Measurement and possible function of GnRH in cerebrospinal fluid in ewes. Reprod Suppl. 59:25-39.

Okay, so for now, don't really worry about CSF, because we don't have clear functional data yet, just growing evidence pointing that way. My current work focuses more on the neuroanatomical side of the issue because there is no easy way to do a direct test of function just yet. Though, I'll be starting pilot studies this fall in an attempt to develop a method to test it directly, but don't yet know if it will really work. Now that I have a bit of the anatomical work done, I have a reasonable chance, but still no guarantees. Mostly, I just happen to have the right animal model to even consider it being feasible.

 

[saltydog]

I'll be starting pilot studies this fall in an attempt to develop a method to test it directly, but don't yet know if it will really work. Now that I have a bit of the anatomical work done, I have a reasonable chance, but still no guarantees. Mostly, I just happen to have the right animal model to even consider it being feasible.

Good for you Moonbear and if your competence here is a reflection of such elsewhere I shall have confidence in your success. Thanks for the info.

 

[neurocomp2003]

damn that's some hardcore neuroscience ... to much for me

Doc could you give me a bit more info on those areas you mentioned
"VTA, nucleus accumbens, area postrema"i've heard of the NA but i can't remember where...

as for modelling yeah its a best tedious...but the one thing i realized above all others is that the most important skill you need is the understanding of programming &algorithms not the other theory as much till it comes time for you to use them...i worked with Dr. Sue Becker which was kinda cool because she knows quite a few people Hinton, Neil Burgess,OKeefe, and a few others at UCL. I want to follow in the line of work that's best seen at MIT and CMU/UPitts(robotics and math neurosci) wchih is where i want to do a postdoc if i ever get into grad school

 

[neurocomp2003]

hmm i went to the mouse Library...but is it only atlases without actually being able to highlight areas of the maps? or am i linking to the wrong section?

 

[DocToxyn]

Doc could you give me a bit more info on those areas you mentioned
"VTA, nucleus accumbens, area postrema"i've heard of the NA but i can't remember where...

The VTA, or ventral tegmental area is located in the midbrain, just dorsal to the substantia nigra (in the rodent brain) and is the site of many dopaminergic cells bodies which are, for general purposes, grouped with related projections from the SN to form the mesocortical, mesolimbic and nigrostriatal dopamine systems. It's the projections from the SN that are affected in Parkinson's.

The nucleus accumbens is one of the targets for the aforementioned dopamine systems and plays a large role in reward behaviors and drug addiction, r.e. cocaine. It is located just anterior to the caudate.

Area postrema is one of the zones in the brain where the blood brain barrier is leaky. This region serves to monitor the blood stream for signals which initiate the emetic response in those animals which can regurgitate.

 

[DocToxyn]

No, I wasn't aware of that. I've never been far from an atlas, so never considered looking online for a mouse brain atlas.

Sometimes I'm at my desk and want to check something out and then realize I left all my atlases at the cryotome or other area, so sometimes it comes in handy.

And then a review article on the system I'm currently studying:
Skinner DC, Caraty A. 2002 Measurement and possible function of GnRH in cerebrospinal fluid in ewes. Reprod Suppl. 59:25-39.

I don't have access to the journal, is that sampling technique at all like in vivo microdialysis? We used to use that to monitor "real-time" changes in extracellular dopamine and metabolites in rats and mice following drug challenges, toxicant treatment, etc. Sounds like an interesting topic.

 

[Moonbear]

I don't have access to the journal, is that sampling technique at all like in vivo microdialysis? We used to use that to monitor "real-time" changes in extracellular dopamine and metabolites in rats and mice following drug challenges, toxicant treatment, etc. Sounds like an interesting topic.

Since sheep have a much larger CSF volume to work with, we can directly withdraw CSF via a cannula rather than doing microdialysis (which isn't so feasible in sheep). This gives us a sufficient volume to assay for GnRH, and for that group to characterize very nicely the patterns of GnRH release into CSF simultaneously with sampling of pituitary portal blood, also for GnRH, and jugular blood for LH. The interesting thing that tells me there may actually be some independent release of GnRH into CSF from that released into portal circulation are that you see some GnRH pulses in CSF that do not have a coincident pulse in portal blood or LH pulse in jugular blood (having both measures gives me more confidence there is a real dissociation between CSF and portal GnRH release rather than a detection problem in that portal sample). So, the question is, if there is release into CSF, what is it doing? This is also as much as I want to discuss here; I don't want to get too far off-topic, nor do I want to discuss on the open forum my next set of ideas in answering that question. So far, I've only talked about what's either published by other groups or already in my current grant. If anyone is curious about more, I'd be happy to correspond via PM, but not on the open forum.

 

[neurocomp2003]

thanks doc.
oh and moonbear
infralimbic and prelimbic cortex any chance you can tell me what these parts do that you mentioned in one of the above posts?I found some rat articles that say they are involved in spatial navigation.

 

[Moonbear]

thanks doc.
oh and moonbear
infralimbic and prelimbic cortex any chance you can tell me what these parts do that you mentioned in one of the above posts?I found some rat articles that say they are involved in spatial navigation.

I don't usually focus on the cortex for my work, but the studies I know of involving the prefrontal cortex rarely distinguish between the anterior cingulate, infralimbic and prelimbic (I've seen studies that say they are lesioning cingulate, but when you look at the images of the lesions, they seem to have hit all three), so I personally can't tell you what each one does, other than at least as a group, they are believed to be involved in motivated behaviors: sexual anticipation, drug addiction (or drug seeking) and classical conditioning (e.g., taste aversion). That's all I know of off the top of my head. That's the reason when I first mentioned the prefrontal cortex, I was joking about that being the fun part of the cortex, related to the studies being done on it's role in sex behavior and drug addiction. All three regions are considered part of the limbic system.

DocToxyn may actually be better at answering some of these questions. My research is neuroscience research, but I'm not classically trained as a neuroscientist. I'm more of an endocrinologist/neuroendocrinologist/reproductive biologist who has been dragged into neuroscience kicking and screaming and laughing all the way. So, I've spent the past three years learning my way around the preoptic area and hypothalamus and am now moving toward the brainstem, so while I know some stuff about the cortex as it relates to circuits involved in reproduction and reproductive behavior, that's about all I pay attention to other than knowing what the pretty golgi staining looks like.