Adam A. Wanderer wrote:Thanks! When this comes about, I wonder how it'll speed up solution to the folding problem.
Call me a skeptic, but I don't think it will. This is a totally different class of hardware and while I'm sure it will be very good as solving certain classes of problems, FAH, in it's present form just doesn't "feel" like one of them. In a
recent post somebody asked what a trajectory is and somebody else pointed him to Dan's explanation. FAH computes individual trajectories, one at a time, but most of the time, it's goal is not to understand the endpoint of a completely folded protein, but rather to find all the possible ways the protein might end up in a different final shape than the most probable end-point. We want to know about those exceptional trajectories which lead to a mis-folded protein. Then we can concentrate on a drug that inhibits these paths to a mis-folded protein.
If that hardware is somewhat like the human brain (and I suspect it is), it should be able to find the shape of the most probable end-point much more quickly than FAH can, but it probably wouldn't be good at finding (all of) the unusual shape(s) that might happen in exceptional cases. If it does happen to find one of them, it probably won't know what's different about that trajectory that causes it to progress to an exceptional shape.
Maybe by the time the hardware is available for home commuters FAH will be asking different questions that are formulated in different ways that are amenable to being solved by that hardware, but it won't be FAH in it's present form. Oh, and in it's present form, FAH can also help science search for a better understanding of things OTHER THAN mis-folded proteins. Maybe that's a good example of analyses that CAN be helped.