FAH really doing anything?

[Methal]

After nearly 10 years of thousands to millions of computers folding you'd think we'd have folded everything to fold including the laundry.

whats come of all this? is there any really tangible evidence that my computer running FAH at my expense (at times a lot) has done the world any good?

I'm not trying to anger anyone, even though i'm sure someone's blood pressure is going through the roof while reading this. I would just like to know. I understand the science behind it, but what of the progress its making? Is this just a game now?

[k1wi]

http://folding.stanford.edu/English/Papers

[Aardvark]

Thank you K1wi for that post.

I think anyone that is interested can scan through that listing and by reading the Summaries and Abstracts they will gain an appreciation for what FAH is helping accomplish.....

[ihaque]

Hi Methal,

After nearly 10 years of thousands to millions of computers folding you'd think we'd have folded everything to fold including the laundry.

Your position is perfectly reasonable. Fortunately for those of us still working in the field, it's not true . There are, in fact, two distinct questions related to folding, both of which are still considered open problems.

The first question is, if you know the sequence of a given protein (the amino acids that make it up), is it possible to determine the final structure into which it will fold? A huge amount of progress has been made on this problem, and for some classes of proteins, very good methods exist to construct so-called "homology models", which model the structure of a protein based on similar known structures. However, for many classes of proteins, for which no good structures exist, it's very much open. Folding@home runs some projects along these lines (in particular, several that have been run by vvoelz).

The second question is a bit more subtle. Let's say that we know the structure that a protein eventually reaches; the question is, how does it actually reach that structure? This question is of obvious academic interest; it also has applicability: misfolding (folding down the wrong pathway or to the wrong structure) is implicated in several diseases. Furthermore, protein folding can be considered one form of molecular self-assembly, which is a particularly interesting topic in nanotechnology - how can we get complicated structures to assemble themselves on the nanoscale (since they're too small to manually assemble)? A huge amount of the work Folding@home does is geared towards answering this question, and it's led to some pretty important shifts in our understanding of how proteins fold. In particular, check out Vijay's recent blog post: http://folding.typepad.com/news/2010/03 ... -team.html

[codysluder]

From a non-scientific perspective, I'd think that there's one more question, but I guess you'd include it in the second one.

If the answer to question 1 is this sequence produces this final structure, then how can there ever be anything called "mis-folding"? It seems obvious to me that if 98% of the time the normal shape is produced and 2% of the time people develop a life-threatening disease, somebody needs to know WHY, before they can attempt to figure out what to do about it.

[Methal]

Looks like i'll have to do some reading up..

I do have one more important question....well a couple more

Right now my i7 laptop is working on an a1 (whatever that is....?!) with roughly 20% of my CPU give or take 5-10%

what is it doing? this processor can do billions, if not thousands of billions of operations per second. Why does it take so long to do what its doing with whatever this thing is?

I know there are millions upon millions of different ways one protein can fold. If its a simple matter of covalent, or hydrogen bonds, shouldn't folding one protein take a computer like this one half a second to fold it every possible way?

and if diseases are created by a protein miss-folding, why don't they take the miss-folded protein and investigate ways to refold it? introductions of amino acids that break down the broken proteins or something. (or little nano bots with lazers....i'm sure that would be just as easy =/)

I guess I don't know as much as I think i do. If someone could be as in depth as possible, or at least point me in the right direction to get answers i'd be happier than a redneck with a hot cousin.

[ihaque]

From a non-scientific perspective, I'd think that there's one more question, but I guess you'd include it in the second one.

If the answer to question 1 is this sequence produces this final structure, then how can there ever be anything called "mis-folding"? It seems obvious to me that if 98% of the time the normal shape is produced and 2% of the time people develop a life-threatening disease, somebody needs to know WHY, before they can attempt to figure out what to do about it.

Well, a partial answer to your question is that there's more subtlety to the problem than question 1 let on. It's not always true for every protein that there is exactly one ending structure that is stable, and that that structure is dramatically better than any other one. It turns out to not always be the case that if you denature (unfold) a protein, that it will automatically pop back into its normal shape. While it happens a lot of the time, sometimes proteins will enter a misfolded state, which may not be optimal in a global sense, but is a locally-stable structure. Biology actually has developed an entire mechanism for dealing with this, called "chaperone" proteins, whose role is to help unfolded or misfolded proteins fold into their correct shape.

Even barring consideration of chaperones, proteins sometimes find dramatically different structures that are also stable. In particular, many proteins, when put together in high concentration, will form a very different structure known as an amyloid fibril that consists of an aggregation of the protein. You may have heard the term as regards Alzheimer's; that's an amyloid formed from one particular protein, but many other kinds of proteins can misfold and form these aggregates.

Finally, in some cases, the misfolding is triggered by an actual change in the sequence of the protein. A good example of this is cystic fibrosis. This disease is caused by a mutation in CFTR, a membrane protein that acts as a channel for chloride ions. One of the more common mutations causing cystic fibrosis is the delta-F502 mutation, in which one amino acid (a phenylalanine, at position 502) is missing. This causes a change in the channel that prevents it from being properly exported to the cell membrane on production.

The last case is one that I think falls more neatly into the first question I outline in my earlier post. The latter two definitely fall into the second question, and there's been quite a lot of work done in the lab on the mechanisms behind protein aggregation and chaperone/chaperonin activity.

[ihaque]

I do have one more important question.

Right now my i7 laptop is working on an a1 (whatever that is....?!) with roughly 20% of my CPU give or take 5-10%

what is it doing? this processor can do trillions, if not thousands of trillions of operations per second. Why does it take so long to do what its doing with whatever this thing is?

I know there are millions upon millions of different ways one protein can fold. If its a simple matter of covalent, or hydrogen bonds, shouldn't folding one protein take a computer like this one half a second to fold it every possible way?

Both those numbers are off by many orders of magnitude. Each core of your CPU is capable (at a theoretical peak) of doing a few billions of operations per second. If we're being generous with those estimates, and said you had 8 cores, it would at most be around 100 billion operations/sec (100 GFLOPs, to use the technical unit - giga-floating point operations/sec). Realistic throughput is much less than that, but even that is much less than a trillion.

The number of possible conformations of a protein is absolutely astronomical, on the other hand - millions upon millions doesn't begin to cover it. I've heard estimates as high as 10^143 possible conformations, and as low as 10^60. Let's be very generous and assume it's the square-root of that: 10^30. Let's further say that we can evaluate each conformation with just one operation on our theoretical 8-core machine, and that we have about 100,000 of those machines. That's 10^30 / (100,000 * 100*10^9) sec = ~315 years per protein. Now, in reality, the total conformational space of even a single reasonable protein is probably larger than 10^30, no processor can evaluate 100 billion conformations per second, and we certainly don't have 100,000 of them.

Algorithms are used that are considerably more clever than just enumerating every state, but going through each step is still fairly computationally expensive.

and if diseases are created by a protein miss-folding, why don't they take the miss-folded protein and investigate ways to refold it? introductions of amino acids that break down the broken proteins or something. (or little nano bots with lazers....i'm sure that would be just as easy =/)

There are groups looking at methods for protein refolding, but it's a hard problem when the question of how proteins actually get into that incorrect state isn't well understood.

Reading the Wikipedia articles on protein folding, the Levinthal paradox, and molecular dynamics would not be a bad place to start.

[7im]

Hi Methal,

After nearly 10 years of thousands to millions of computers folding you'd think we'd have folded everything to fold including the laundry.

Your position is perfectly reasonable. Fortunately for those of us still working in the field, it's not true . There are, in fact, two distinct questions related to folding, both of which are still considered open problems.

Sorry to put you on the spot with these questions... I think us non-scientific types want a simpler answer. Considering this question is still open, and considering that fah has been running for 10 years, how much has has fah been able to close up on this still open question? 1%? 10%? What's your best guess?

P.S. I really appreciate your answers above.