I was invited the University of Washington today and met with Zoran Popović, Firas Khatib (Beta_Helix), and David Baker. We discussed a number of things in the time I had there. One thing that came up with David Baker was about how to improve the information about foldit that is available. I told him that I'd at least try to form the list of questions that need answering.
This is a chance to improve much of the information available on the wiki. So here is your chance to ask questions about things related to the biology in foldit. You ask it, and we'll try to get some answers out of the foldit guys.
Just to give you an example of the kind of question I think we might get answered:
What does rebuild actually do? As I understand it, it looks at a database of segments from proteins with known native structures and tries to find segments that fit the sequence of amino acids and the endpoints of the rebuild. How exactly does that work? Where does that database come from? Who maintains it? Is it being added to as more natives are determined? Is it hardcoded into the program or downloaded from the server (i.e. can it change frequently or at all?)? Is this even the correct understanding?
Ask away, and we'll try to get the questions answered using words that a non-biologist, like me, can understand.
If I isolate five segments with frozen segments and launch a rebuild, stop the rebuild, undo to the starting point of the rebuild, and restart the rebuild, will it just repeat all the positions from the initial run? Or will it generate a different sequence of positions?
What exactly is one iteration when perfoming wiggle/shake?
Sometimes, a wiggle using one iteration does much, sometimes almost nothing.
What exactly does this tool do? I know it can change the identity of (mutable) sidechains from one amino acid to another. But I have seen it raise my score even when no amino acids are changed. I would like to have greater knowledge so it can help my intution in the use of tools - I would like to know how they all work.
Try it and see. You can use the undo graph to make a record of the scores of the different poses. Then repeat from the start point. I suspect you will find that the scores are different - hence the poses are different. I know that you can run a rebuild script twice on the same starting save and get two different results.
I'd like some more info on what this tool does. It seems to have some random component, since you can run a rebuild script twice on the same starting save and get different results. It seems to be "clever" in that if you leave in structure and add bands on sidechains, it will try to give poses that preserve your structure and shrink your bands. But I suspect that when run on a selection that is all loops, it just gives random poses that start and end at the endpoints (roughly speaking: of course the end segments and those adjacent bend and twist, but the point is that the rebuilt shape connects up with the protein). I'd like to know if it is just working completely randomly within these constraints, or if when you run it for a long time and let it generate many poses it is applying some algorithm to search for some shape (I suspect not, but I'd like to know). Similarly, I'd like to know if it takes any account of sidechain positions when generating poses (again I suspect not). It would be nice to have some more control over the rebuild poses since at present there is no guarantee the tool will ever give you a pose in the right shape you are looking for.
Is the importance of FoldIt merely conceptual for chemists, or can the FoldIt results be used in understanding of actual proteins? If the latter, how and how can FoldIt players provide more useful feedback to chemists and vice versa? Does it matter how many moves we use in folding proteins and whether we use scripts? Do chemists prefer many different solutions or just the highest-scoring fold, and why?
Ask if there is any commitment to improving the recipe editor. Particularly the Linux version without the copy/past facility is onerous and nearly useless.
What are the uses for the secondary structure in Foldit. I would guess: to make a difference between backbone wiggle and wiggle all. There are other, probably correct suggestions in the Wiki, but really what does it do? In actual proteins these structures do have different behaviour. For example the alpha helix tends to destabilise because of charge build up. Is this effect incorporated in the calculations, and if it is, is it dependent on the choice of the secondary structure?
Game related: what does clashing importance do? I found al lot of hits on this in the Wiki but an explanation did not immediately appear.
In general I would suggest to extend the descriptions of the Foldit functions in Lua. It was not apparent to me that the select function can select multiple ranges by applying it more than once. This good behaviour of cause, but it surprised me when I found out. I think it should be in the documentation…
I agree with apple.muncy, Linux without the copy paste facility is worthless…just my 2 cents!
