You are all doing great things on the alignment puzzles! Here are a few tips to keep in mind:
1) The biggest errors in the starting alignment will often be in strands at the edge of sheets. It is very hard for automatic methods to align these
properly. You may want to try shifting the alignments up and down several residues, allowing the loops before and after the beta strand to change in length
to allow for the sliding of the sequence through the strand.
2) Loop regions that are close to a sheet in a starting model may in fact be strands, so you might try converting the loop to a strand and making it pair in the sheet. A beautiful example of this is the miniCASP 14 puzzle. The highest scoring FoldIt solution converted a loop near a sheet to a strand, and totally nailed the correct solution. I found this extremely impressive!
3) Loop regions often vary quite a bit during evolution. Look carefully at the sequence alignment in the loop regions, and try shifting segments right and left where the alignment is poor. You will obviously need to spend quite a bit of time trying to optimize regions that are not aligned.
4) Regions where the sequence alignment is very good are likely to be correct, and it is probably wise to leave these more or less alone.
Congratulations on your work so far, and we are expecting great things during CASP which starts in less than a month!
Below is the top scoring Foldit solution (in green) for Mini-CASP 14 - Last Chance, the template that was used (in red), and the native structure which was released yesterday (in blue):
Is that some gap in terminology between what David is talking about and what is used in the game?
In the game, we have loop, helix and sheet as structures, and David is talking about sheets, loops and strands.
Which is which?
See http://en.wikipedia.org/wiki/Beta_sheet#Nomenclature
"In the most common usage, β strand refers to a single continuous stretch of amino acids adopting an extended conformation and involved in backbone hydrogen bonds; by contrast, a β sheet refers to an assembly of such strands that are hydrogen-bonded to each other."
Besides, I was wondering about the native(!) curly β strand in the bottom left corner of the picture. I'd think this is a random coil (loop), not a strand.
Thank you for clearing that up, Madde.
Brick you are correct that in the game we are always talking about:
helices
loops (or coils)
sheets (which are made up of strands)
David was focusing on beta-sheets and loops because those are often the biggest errors with alignments (which you all have probably noticed already), whereas Rosetta does not have as many problems when it comes to helices.
Madde, to answer your question about the bottom left corner of the native… that is actually PyMol's secondary structure assignment thinking it's a helix!
You can maybe see it more clearly here:
http://www.rcsb.org/pdb/explore/images.do?structureId=3LWC
The secondary structure in the above cartoons are even worse, with strands curving like boomerangs!
I completely agree with you, that looks more like a loop than any strand or helix.
Unfortunately, proteins don't actually look like these cartoon representations, we just use them because (most of the time) it's easier to visualize the fold of a protein in cartoon form.
