Folding@Home Press FAQ

We've compiled many of the questions asked by press and answered by Prof. Pande into a FAQ below.

How did this all come together?

The Folding@Home project has been running for several years. How did it begin?

In 1999, I started my position a professor at Stanford University, heading a research group. In order to push the envelope on what could be accomplished by my research group, it became clear that we would need to make a major push in computational power to perform our research of interest.

One of the first challenges was to develop a way to efficiently use distributed computing for our research. Not all problems are well-suited to distributed computing and ours was on the surface, not particularly well suited. However, after the development of novel algorithms, we discovered a technique which allowed for distributed computing to make significant advances in what we do.

What is distributed computing?

Even back then, the calculations we wanted to do would take about a million days on a fast processor. So it seemed natural that we might be able to get it done in 10 days, if we had access to 100,000 processors. So in using distributed computing, we could run pieces of the simulation through networked computers to speed up the results.

Why don't you just use a supercomputer?

One typically holds 5,000 processors but each is often surprisingly slower than a fast processor in a PC these days. Also you never get the supercomputer to use for yourself, you share the processors. So the power in terms of raw performance in comparison to what we can do in folding@home, is about one hundred times slower.

How did you come to be working with Sony? Did you approach them, or was it the other way round?

Sony approached us (some people at Sony were interested in a collaboration before they chatted with us), but we had been looking to work with them too – it’s easier for them to contact me than vice versa.

What does FAH do?

Most people will assume the research you’re carrying out is too complicated to fully understand. In the interests of getting people interested in the project, is it possible to explain it in layman’s terms?

Our work centers around proteins. Thus, it is natural to ask “what are proteins and why do they ‘fold’?” Proteins are biology's workhorses -- its "nanomachines." Before proteins can carry out their biochemical function, they remarkably assemble themselves, or "fold." The process of protein folding, while critical and fundamental to virtually all of biology, remains a mystery. Moreover, perhaps not surprisingly, when proteins do not fold correctly (i.e. "misfold"), there can be serious effects, including many well known diseases, such as Alzheimer's, Mad Cow (BSE), CJD, ALS, and Parkinson's disease.

With that said, what does Folding@Home do?

Folding@Home is a distributed computing project which studies protein folding, misfolding, aggregation, and related diseases. We use novel computational methods and large scale distributed computing, to simulate timescales thousands to millions of times longer than previously achieved. This has allowed us to accurately simulate folding for the first time, and to now direct our approach to examine folding related disease.

Are there any major disadvantages to conducting the research in this way?

There are always pros and cons of different styles of research. If someone donated several billion dollars to Stanford for computing resources, we would be able to build a resource comparable to Folding@home. However, that’s not possible. So, within a more realistic budget, distributed computing is the most cost efficient means to build a supercomputer of this enormous scale.

What benefits can these simulations of protein folding bring?

With simulations, one can study aspects of folding and misfolding (and related disease) that one could never see with just experiment alone. Simulations won’t replace experiment, but can be a critically useful tool to go beyond what one could solely do in the lab. We are combining our simulation predictions with laboratory tests (either done in my lab or in collaborators). Working together, we can greatly push the boundary of what used to be considered to be impossible, even just a year or two ago.

How accurate are Folding@home’s predictions?

We are constantly testing Folding@home’s results for their accuracy, as this is a critical aspect of our work – to make predictions relevant for experiment.

Is there any cost to the donors?

We don't charge for our software or anything like that. Donors must buy their computers, etc, but we don't charge them anything -- they are donating a great resource to us, so there's no reason to ask them for more.

Folding@home's efforts in computational drug design

How about your work in drug design?

As our work matures, we have been looking to take the next steps involved in computational drug design. Our work here focuses on building a complete pipeline, starting with new methods for docking (work by Kim Branson in the Pande lab) to filter a million molecule library down to 100 to 1000 molecules and then use free energy methods (work by Guha Jayachandran, Michael Shirts, and John Chodera in the Pande lab) to filter this down to 10 to 100 molecules to assay experimentally (currently done by Kim Branson in the Pande lab and in collaboration with other labs as well).

How is FAH being used to implement computer-aided drug design?

The two critical issues in computational drug design is speed and accuracy. usually, one has to trade one for the other and use a computationally efficient method like docking, at the cost of some accuracy, or use a computationally demanding method like free energy calculations, but at the cost of low throughput. With distributed computing and a set of complementary methods (both docking and free energy calculations), we can have it both ways -- with the goals of high throughput and high accuracy. Distributed computing is a key aspect to this, as it allows us to do calculations otherwise impossible.

Specific examples are great--and if you have any pictures to accompany these examples, that would be great too.

We have some work published in this area (see the FKBP work done by Shirts and Jayachandran), but much of what we're most excited about has not been published yet. Our areas of interest in computational drug design are largely in 3 areas: cytokine-cytokine receptor interaction inhibition, development of novel chaperone inhibitors, and novel antibiotics to target the ribosome.

What have you done so far?

What benefits can these simulations of protein folding bring?

With simulations, one can study aspects of folding and misfolding (and related disease) that one could never see with just experiment alone. Simulations won’t replace experiment, but can be a critically useful tool to go beyond what one could solely do in the lab. We are combining our simulation predictions with laboratory tests (either done in my lab or in collaborators). Working together, we can greatly push the boundary of what used to be considered to be impossible, even just a year or two ago.

How can you be sure that your simulation is accurate?

We are constantly testing Folding@home’s results for their accuracy, as this is a critical aspect of our work – to make predictions relevant for experiment.

And if it is, what will it tell you, and what actions can you take as a result?

We have a tight coupling between simulations and experiments. New simulations, lead to new experiments, which lead to new simulations. With this iterative process, we hope to first gain a better understanding of folding and misfolding, and then apply this understanding to the development of novel drugs and other types of therapeutics.

What advances have you made in your research since the Folding@Home project began?

We've been able to push the boundary for simulations by several orders of magnitude and used it to make advances in understanding protein folding, misfolding, and related disease. Check out http://folding.stanford.edu/awards.html for what others have said about us.

How long are you planning on running Folding@Home. Lets look ahead five years - what do you hope will have been achieved?

The goal of the first 5 years was to make a significant advance in understanding folding. My goal in the next 5 years is to have made a significant advance in our understanding of misfolding and related disease (esp Alzheimer's disease). We have 4 years left to do that and I feel that we are very much on track.

Could you summarize some of the key papers which have resulted from FAH?

This is always tough to do (as it's like asking someone with ~50 kids which ones are his favorites), but here's a sampling. See http://folding.stanford.edu/English/Papers for more details and the paper #'s I'm refering to.

de novo simulations of protein folding with quantitative agreement with experiment While paper #1 (Shirts and Pande, Science, 2000) got the ball rolling, paper #8 (Snow et al, Nature, 2002) was important since it was the first time that experiment and simulation could really match in this sort of quantitative fashion. It was a test of many aspects of FAH and turned out quite well quantitatively. Paper #17 is another good example of this, where we compared to multiple experimental methods. This early work has been followed up by numerous works after to better understand folding, including folding in vitro (53, 49, 45, 42, 37, 35, 33, 24, 23, 22, 19, etc) and models of in vivo (#50, #36).

New drug design methodology We have also been pushing the boundaries of what can do with computational drug design (method in paper #29, results in paper #31 and #43, where we showed that our methods were very accurate, for a target of pharmaceutical relevance)

New methodology to simulate folding on distributed networks We have also had major efforts to further enhance our methods to push FAH to do more and more. This includes papers 54, 49, 46, 40, 32, 27, 26, 19, etc.

Applications to disease Most of the exciting work is still under peer review (I think it always feels like that for scientists, as it takes ~1 year for review/publication), however some highlights that are already out include our work on cancer (papers #39 and #20) and lipid fusion, relevant for viral infection (papers #41, #47, #51)

Science background

What are proteins and why do they need to fold?

They are nature's nanomachines. Proteins are the molecules in the body that it uses to get everything done. They act as catalysts to speed up chemical bonds that might take a billion years through other types of biological machinery. So whenever something needs to get done in biology, odds are, proteins are at work. They all have different functions, some are like scissors, some bring bonds together. If you think about the use of building nanomachines today, biology solved this millions of years ago. It amazes how well this works in the body.

What happens when the process does not work?

When proteins do not fold correctly, that is when trouble occurs. Lots of ways this can occur. A human can be missing the protein, like in cystic fibrosis. What is more prevalent is there are a whole class of diseases that were thought not to be related to one another, such as Mad Cow, Alzheimer's, Pakinsons, types of cancer and ALS, that are related to protein misfolding. And the misassembly not only makes the molecules toxic and dangerous but the misfolded proteins recruit proteins behaving themselves and they are now toxic.

Latest breakthroughs in the field of biochemistry?

The connection between biology and nanotechnology is very exciting and for decades it has been a dream to be able to rationally design drugs. What I mean is you design a bridge and cars go over it and it works. The way we design a drug, using the bridge analogy, we would create a bridge, send some rats over it and if it they survive, we try humans who really want to cross the bridge. It is still very much empirical The dream is to design molecules the way we design macroscopic objects. Sounds easy, but being so small you must be really, really accurate and it is very computationally demanding.

Folding@home on the PS3

How did the PlayStation 3 get involved?

Even with our Folding@home program for PCs, there are certain calculations, even with the amount of computers we have, that still take 18 months to run. When I saw the specs of the PS3 come out I knew it could have a lot of potential power for the type of simulations. We worked with Sony and the early tests showed better than a 20 fold speed up.

What's in it for the gamer?

It's a visualization of the dynamics of the simulation you are seeing in real time. We worked with Sony to make the simulations visually appealing and more in line with what a gamer might see to make them interested in the program. We also have teams of PS3 users, a concept very old to distributed computing, that is sort of a competition. We keep track of how many computations a user's computer has done and offer a leader board. We also keep track of team points.

What are your goals with the program?

We can handle about two million PS3s right now but only about 30,000 are folding. We hope to be able to get more PS3 owners involved and make some exciting experimental predictions that bear out. It's hard to know long it will take to come up with therapeutics for diseases. We could be close or a hundred years away but distributing computing will help speed up the process.

Is there anything about the PlayStation3 that makes it especially well-suited to the task, or could the project feasibly run on the Xbox 360 too? Do you have any plans to further expand the project?

There are several aspects which makes the PS3 well suited for this task. First, it is powerful: its main processor – the Cell Processor – is very powerful. In fact, we get a 20x speed increase over PC’s. That’s not 20%, but 20x, i.e. a 2,000% increase over a typical PC. That sort of speed can’t be found on PC’s or on the xbox’s central processor.

Second, a console environment is very uniform, which makes it easier for us to support and easier for people to run Folding@home. Indeed, the PS3 version requires just 1 click to run. One could never do that on a PC version.

Finally, the graphics/visualization of the PS3 isn’t something we could easily do on a PC. Game consoles push the envelope of visualization for games, and it’s great to take advantage of this for Folding@home.