A protein folding game does not have to necessarily be about protein prediction from amino acid sequences with unknown tertiary structures.
Fold.it is not about protein prediction from amino acid sequences with unknown tertiary structures. It also includes folding Amino acid sequences with known tertiary structures as well, which is what is done in the Revisiting puzzles.
It is about using the 3-D pattern-recognizing capabilities of the human brain to optimize the automated algorithmic computations to predict protein folding, to solve the problems present in the automated algorithmic protein design algorithms that have been developed so far. It says so in the "About" page.
The thing is, as I think, the Foldit player's predicted solutions of prediction puzzles seems not that diverse and as "creative".
The process taken by experiences Veteran Fold.it players is mainly fixed and un-diverse,
by just running the primary structure through PSPIRED,
then folding the SS so the hydrophobic amino acids are inside & necessarily pairing any viable sheets,
and then running recipes to squeeze out the rest of the points.
That is all. It is not that hard.
It seems as if predicting the tertiary structures from amino acid sequences is fairly easy and straightforward after all.
I think the developers of the protein folding algorithms got the hang of it already, so it seems that there's going to be a protein folding algorithm that predicts the structure of unknown proteins from a given amino acid sequence soon.
On the other hand, the automated algorithmic protein folding algorithms that have been developed seem to be inefficient with designing proteins that can bind to other substances from scratch, and that it seems hard for a computer to bring amino acids and Secondary structures in specific, fitting places on the interface of the receptor, especially since computers are very dumb and can't see like we do.
So, it seems that the Foldit team is trying to use the human brain, which is used to a 3-D environment, unlike computers, including depth, shadow, proximity, groups, etc., to fold proteins that can bind to receptors.
Also, the revisiting puzzles are to see how much more the human brains from Foldit players can predict the tertiary structures from a given amino acid sequence, along with the testing of the accommodation of the new capabilities of protein folding actions.
This means, that they are testing to see, "How much have the results of revisiting puzzles been improved from before, with the new capabilities? ". So, Foldit players are technically predicting the structure of a protein from a given Amino Acid sequence.
There is a place for prediction work on the puzzle where the top 10/20/50 solutions are equal and just copied from the PDB. It seems that it is where Foldit scientists don't just look at the top solutions, if they are all the same.
If they are all the same, then it's counted as just 1 solution of course. It's logic.
It seems that the Foldit scientists take into consideration other lower-scoring solutions, and take into consideration the duration and sequence of moves, and hypothesized factors of reasoning and progress as well, along with the progress made by the difference from the new protein folding capabilities.
In my opinion, I think that the revisiting puzzles are to see how many more solutions are copy-cats from before, so they can distinguish the copy-caters from the rest.
That seems to be one of the differences that the Foldit team is trying to get ahold of, and that each Revisiting puzzle seems to be one data point, that is categorized on a variety of factors based on the properties of each protein in each Revising puzzle. This may be why there are so many revisiting puzzles, which is for accuracy.
Plus, like bkoep said, the Revisiting puzzles do provide a good puzzle level for players who are bored with the beginner puzzles but are not that experienced enough to play the advanced puzzles that well. So, that type of players might learn and do well in the Revisiting puzzles.
That protein prediction competition, CASP-14, is just prediction, as you say it is. It seems to be not helpful. I don't care if it's popular, or a competition, as long as it has no purpose for me. Just like the Tides Pods Challenge. Sure, you can do the challenge, but in the long run, the challenge is a waste of time, with no true, helpful, important, necessary purpose.
Foldit players just predict a tertiary structure from a amino acid sequence to that puzzle, but it seems that now, what science will you get from that? It seems to be relatively straightforward to predict a puzzle from a given amino acid sequence. But it seems to not be as relatively straightforward to predict a protein that a receptor will bind to, given just the tertiary structure of the receptor, without an amino acid sequence to start with. That seems to be one of the protein design problems that the user, bkoep, seems to be talking about.
Also, a .it domain doesn't necessarily have to be from Italy. It's just used by Italians the most, according to a Wikipedia article about the .it domain.
According to a Wikipedia article about the .it domain, "Because it is also the English word it, and many words end with -it, this can commonly used in the construction of domain hacks, such as play.it (now owned by CBS Radio), write.it, make.it or rabb.it." So the .it domain name is not about Italy, it refers to folding the primary amino acid sequence , which is present in all of the fold.it puzzles, including tutorials and contests and science puzzles. So, the .it domain name in Fold.it is just a domain hack, with, according to https://en.wikipedia.org/wiki/Domain_hack ,"In this context, the word hack denotes a clever trick (as in programming), not an exploit or break-in (as in security).".
So, since every puzzle in Fold.it involves the folding of the primary amino acid sequence, Fold.it is a protein folding game.
Quod Erat Demonstrandum :)
From, donuts554.
