Today, I’m going to post a nice summary of one major facet of FAH, the study of protein-ligand binding. This issue is critical for computational drug design, and brings together many parts of what we do with FAH. This summary was written by Peter Kasson and I’m going to paste it verbatim, since I think he did a nice job with it.
We’re often interested in comparing things–predicting a known difference is a good way to test our methods. Then, once we’re pretty confident that things work, we want to predict ways to change the way proteins interact. Changing a system in a defined way is both a good tool for biological insight and the basis for a lot of medical treatments.
In this particular case, we’re interested in the "selectivity" of ligand binding by a protein: the protein is known to bind one small molecule ("ligand") much better than another. So 3903/3905 is a pair of projects comparing the protein-small molecule interactions, one project for each small molecule. 3906/3907 are the essential "control" projects that tells us how different the small molecules are in the way that they interact with the water around them. When we combine these pairs of projects, we can then calculate the difference between the "bound" state in which the protein is interacting with the small molecule and the "unbound" state in which the protein and small molecule each just interact with the environment. There are experimental data on the difference between these two small molecules; if we get good results here, we’ll go on to test a number of interactions that aren’t experimentally known yet.
The trick is that some protein-small molecule interactions are quite easy to test experimentally and some are very hard. Using Folding@Home, they are all moderate in difficulty. So if we can validate our methods on the "easy" ones, we can then predict the "hard" ones. Usually only a small fraction of protein-small molecule interactions that one tests turn out to be important. For the ones we predict are really interesting, we can do the experiment. But Folding@Home allows us to skip the experiments that are both hard and likely not to be interesting. This sounds simple, but it can be very powerful in trying to understand the underlying biology.
We also hope that it can help us understand a number of diseases and drug interactions better. For instance, a number of diseases or "failures" of medical therapy are due to mutations in proteins (changes to their amino acid sequence). We would like to understand how these mutations affect the interactions between the protein and the drug. If we can understand that, we can help suggest ways to improve the drug therapy. This is obviously a large, hard problem, but we think that Folding@Home can play an important role here.