Hello,
For CASP targets that are refinement targets, we usually are given the GDT from the native. It would be helpful to have some gauge of what the GDT is of the structures we are producing to the starting structure we were given (so in some cases to try to rebuild regions but not deviate too far when the start GDT is very high).
Thanks for your consideration.
Interesting.
On the most recent refinement, which had only 55% agreement, I loaded the original structure as a guide to see what had changed on initial wiggle. Perhaps you could provide the GDT to the guide. Might be useful for QTTN puzzles too.
This is part of the inspiration of exploration puzzles. In a lot of puzzles we have a general idea of how far off a given model is from the native, and exploration puzzles attempt to capitalize on this by granting a score multiplier when you're close to the correct distance from the start. Unfortunately, so far these puzzles haven't worked out that well, but perhaps it would be useful to include the distance purely for reference, instead of in the score term.
@ brow42 – that's actually what I meant. The GDT to the guide is the useful thing to keep track of. Thanks for making it clearer. @jflat06 I agree, just having the value for reference would be useful. Whether a multiplier effect should be incorporated I would leave that to those in charge to think about, but I think having the GDT to the guide/starting structure for refinement somewhere would be really helpful.
+1 It would be good to have a clue how much we have changed from the template (or from the guide if we load template as guide). In general we seem to change too much on refinement puzzles where only certain areas are supposed to change; I don't know how we do on the ones where no area is specified. Let the players see what you want us to optimize. Some will still go for highest score, but at least some will try for best CASP submission.
This should be disabled in QTTN, since this is a fairly monotonic function one could climb in a recipe.