Setting assignment

As an adjunct to this question, since CASP uses GDT scores to measure proximity to native, would it be possible for foldit to also display GDT scores in addition to current game score? Also, what are the differences between game score and GDT?

Here is a thread also dealing with GDT:
http://fold.it/portal/node/988108

@dimension9:
As far as I know, the puzzle score is the ENERGY level of the proteine (lower energy level is higher score).

As worker for electronics, I imagine this as potential difference of clouds.
If clouds rub together (or cloud with different layers of air near mountains or the sea), they get a different energy potential,
because electrons swap places from cloud to cloud or cloud to air.

An electrical storm (flashs) brings the potential between clouds back to one level, so the potential difference between the clouds is small again (but not zero).

So, you can imagine that a puzzle state at start is before an electrical storm and after all those shakes, wiggles & co it is like after the storm.

The GDT measures the SPATIAL distance difference between the current postion of each puzzle segment according to the position of the same segment in the native structure.

Correct me if I'm wrong.

regards,
Alex

There are two different GDT numbers:

The common used one is GDT_TS (Global Distance Test Total Score), which will be measured
GDT_TS = (GDT_P1 + GDT_P2 + GDT_P4 + GDT_P8) /4
at which GDT_Px is the percentage of Atoms within a distance to the "native" ones of x Ångström

And there is a more accurate one, the GDT_HA (GDT High Accuracy), which will be measured
GDT_HA = (GDT_P0.5 + GDT_P1 + GDT_P2 + GDT_P4) /4

The problem with an experimental solved "native structure" is, that it isn't the real native structure even though it is very close to it. For example, the "native structure" of CASP9-target T0520 as seen in the PDB has only a resolution of 2.6 Ångström (T0523= 1.7Å, T0525= 1.3Å, T0527= 1.49Å,...)

So, when we find a solution with a higher score (= lower energy level) than the "native structure", it is indeed possible that our solution is nearer to the real native than the experimental solved one.
It is even possible that we'll find the exact native structure, but it wouldn't have a GDT of 100 because of the inaccuracies in the "native structure".

Beta_helix, please correct me if there are errors in my argument.

How important is it to science to get the fold exactly right. If you predict the protein will fold to X and are off by two angstroms, will that cause the proteins that you predict will bind to it to be wrong. How good is "good enough"? I'm sure the answer, like most things is, "it depends", but I'd like to hear what it depends on.