A simple question about folding


Feb 28, 2007
OK, kinda' bear with me fellow folders. This question has been sticking in my "craw" since I was reading the thread "How to seedup molecular modelling on GPU?" I was cool with the title because I figured it was just some medical "mumbo jumbo" understandable only by a person or persons involved in the medical field or a closely related side endeavor (which I'm not and don't really want to be) I was cruising on down the read (post) and I ran into a post by a distinguished [H]orde folder What really confused the hell out of me was at the beginning the dude/dudette OP was talking about "possibilities to reach hi performance in molecular dynamics on NVIDIA CUDA technology", whatever the frigg that is and then I get to {H]orde folders post and he says quote/"This is how it works in the real world: medicinal chemist asks for help from computational group....gets answer....tries this out in the lab....doesn't really work. Computational guy refines the model...says try this now....still doesn't work. Medicinal chemist goes and tries something different it works and the project is over"/unquote :confused:

I understand we are dealing with a "trial and error" type deal and that it may take a while to help find a cure. Hell, we might just help Stanford find a cure by accident. What it appeared to me was if you're not a "medical chemist" you're just spinin' your wheels. Is that what Vjay is, is that what he's teaching? I'm sorry if this post is confusing because that's what I am, totally confused. I lost my real mother to cancer many years ago and I would do anything in my power to help find a cure for cancer. I'm about 60 yrs old and I think it's kind of late to strike out in the field of "medical chemist" (it has zero interest on my part anyway) ;)

Would someone that understands folding at home or is in confidence with the medical folding Gods Please post a summary on the Stanford F@H project and Please make the summary understandable by a person of average intelligence, that is in no way connected with any medical field, just people in non medical jobs that see a doctor when they're ill or a computer hobbyist :D (ie a person that uses, likes computers or both, is involved with IT, etc)

Try and get this straight there is NO way I' going to abandon the F@H ship as the above post tries to say "I just needs a simple dimple, uncomplicated, no rhetoric, unbiased explanation of Stanfords F@H and "protein folding". :rolleyes:

Thanks "a' moondoe" (SP) in advance :D

EDIT: Oh yeah, the explanation can't sound like something out of the "Twilight Zone', it has to be half a$$ed believable :D




[H]ard|DCer of the Year 2008
Oct 12, 2004
You posted this in the right time, I just understood exactly what the F@H project do after decrypting what Dan Ensign explained in cumbersome terms. Here's a good explanation in non-scientific terms :

The root of the project is to understand how proteins folder happen and how proteins misfolding can affect the growth of proteins and cause various diseases like Huntingdon, Alzheimer, etc... This is what we call the frame of the project.

Obviously, the best way to do that is to observe the proteins folding directly but there is problems arising from this method : the proteins fold and unfold in a matter of nanoseconds or microseconds, too fast for the naked eye to catch any action. Even observing a single protein behavior (assuming we got the necessary to overcome the short length of the event) doesn't always mean that you will witness a special event and there is billions of proteins to see. This is where my next paragraph will clarify what we do...

To accelerate the protein folding/misfolding events, we run a computer simulation using various types of proteins and solvents (the surrounding where proteins bathe). This is where molecular physics simulation will be involved. This will branche out in a few forks within the project : 1-distributed computing understanding and challenges (for the Computer Engineering students and researchers like Adam Beberg), 2- Molecular simulation challenges to find how to find the best simulation model which can approximate the most closely actual protein folding/unfolding (this is the area for Biophysicists like Dan Ensign) and 3- Studying the results of the simulations to draw out hypothesis and possible conclusions, to be studied with other means. this is for molecular biologists and biochemists like Vijay Pande.

Right now, a big part of the study is to research which way can do simulations the best and with which client. A SMP client can do Explicit solvent model (Explicit in the sense that we also calculate the solvent atoms in the surrounding of the protein model) which is a very accurate but slow method. Another method is with Implicit solvent model (Implicit mean that instead of calculating the solvent atoms individually, it's calculated as a whole with approximations and constants) which is very fast but less accurate. when they are talking about comparing results, this is what they do to see if Implicit solvent models can give results as accurate as Explicit solvent models. If not, what can they do to improve the accuracy. Another thing to do is to see how we can export the Explicit solvent model simulations on GPU2 which is currently impossible.

This is why some of their papers has nothing to do with actual protein folding but about the distributed computing model, simulations methods and even methods to randomize particles as accurately as possible.

Hope this shed a light about what they do ?