There is a problem with foldit as a scientific tool that I can see (and also as a game with a purpose). In the ongoing Grand challange 8 (all hands) good solution like the one in the attached file foldit_1236160851.png becomes superseeded by foldit_1236160936.png were most of the structure is frosen and held together by rubber bonds. The same has probably been done for foldit_1236160965.png where all the coils and beta sheets probably have been removed forcefully by freezing and bending into shape. These structures are not what what we as scientists look for as good solutions. They are just competition solutions. A coil would form more or less directly when the growing peptide chain leaves the ribosome. After that it is not likely to be straightened out and become stabilised by other straightened out coils! It is a much too unlikely event. Suggestions: 1. Make freeze less powerful by only allowing maybe 2 simoultaneous freezes 2. only register points after a wiggle when all freezes and rubber bonds have been removed.
Hi Madde,
1. I know Foldit is not a simulation of the natural folding processes. On the other hand the final solutions should preferrably be something that could possibly have appeared through a natural folding process and not just wierd constructions made possible by the inbuilt rules of Foldit. The game looses too much purpous if it is like that. It becomes just another game. If the game does not become better in rewarding only possible natural solutions I will not recommend it to my student in cell biology in the future.
2. This is obvious
3. OK I know this. But you find high scored shared solutions where rubber bonds and freezes have not been removed. I do not think scores should be registered at all until bonds and freezes are removed and after global wiggle.
I think the key point is that the accounting for scoring should not consider banded or locked configuration's scores. This is a good idea but hard to implement.
Crippling freeze is overkill. One could imagine in nature that a portion of the protein gets "constructed" in the absence of the other portion. I think freeze could in fact be permitting a "somewhat natural" locale for the unlocked portion thus permitting manipulations that create a natural configuration. No proof, but it seems likely or at least possible.
As pointed out freezing creates unnatrual configs as well. So be it.
As to unstable configurations, this is an interesting issue. One has to assume that the solutions being generate are not totally off the wall and are representing things that a good approximations.
I find that the puzzle 127 "all hands" has a great score, however wiggle it after a pull and it settles to about 10 points lower. Some expert opinion as to the merit of this unstable high score would be nice to hear.
Is the lower score the more natural configuration?
If/Does an "unstable" config become more stable after some more manipulations? Meaning that the unstableness is just an intermediate natural state?
Interesting questions.
As i see it there could exist stable but improbable totally off the wall solutions. A protein folds by thermal wiggling and being knocked about by the molecules in the surrounding watery solution. It quickly assumes a stable low energy state where it will spend most of its "working life" in the cell. There might be other lower energy states that would take a protein on average to long time to assume due to the improbability of the chain of events that has to occur for it to happen within the protein lifetime. I am not shure these types of states excists but if they do there would be no evolutionary punishment for a cell to make a protein that has a 50 % probability of assuming a useless very low energy state in a years time. On the other hand we might easily force-fold proteins into these off the wall states using freeze and pulls. The easiest way to see if these off the wall solutions emerges as highly ranked in the game is to select a few proteins containing coils and beta-sheets that have been well characterised by x-ray chrystallography and see how well Fold-it gamers perform. This should be done every time something is changed in the tools or program running the game. I know some relatively short proteins that would be good to test out in this way.
Lol, there can be. I've made them, if only to prove a point about the model and where it needs to go in the future to avoid those mal-predictions, either programmatically or via education. I prefer the latter but that has not been forthcoming and it is unlikely that some programmatic constraint can be put in place to remedy the accidental or purposeful alteration of a protein to a non-realistic solution by some coded response, especially not on an unknown…ok, I could argue with myself over that last statement but I'll stick with it for now for its general consideration…
I believe that the solution to the problem you outlined is one of philosophy more than anything else. If players are trained not to resort to cheap gimics to boost score, then that should go a long way towards stopping the practices that lead to those situations. Conversely, if players are encouraged by other players to "slip under the rope", then it's doubtful that it can ever be stopped.
However, with that said, I would not be at all inclined to fiddle with freeze in that manner.
I agree with some of the other stuff you said as well but I don't have time to respond.
