Mini-CASP 17 Voting Results!

Started by beta_helix

beta_helix Staff Lv 1

The solved structure for Mini-CASP 17 was just released and we have compared the native solution to your votes from the first 'Solution Voting Puzzle'.

Using GDT (Global Distance Test), which is the metric used by the assessors in CASP (the higher the number the closer to the native), you can see that 26% of you selected the Foldit prediction which was the closest to the native:

33% of you selected 'Solution 2' which had a GDT of 44.4
26% of you selected 'Solution 5' which had a GDT of 67.0
18% of you selected 'Solution 3' which had a GDT of 65.1
18% of you selected 'Solution 4' which had a GDT of 65.7
5% of you selected 'Solution 1' which had a GDT of 61.6

We are reposting '272: Mini-CASP 17 - More Alignments' as a Quest to the Native so that you can see the native and see what alignment changes needed to be done in order to get closer to the native state. For example, you'll notice that the template that started with the highest score (top alignment with a score of 127) does not get you closest to the native after a simple threading.

You'll also notice that this Quest to the Native is missing the first 9 residues, the starting extended chain has been trimmed as well as the query sequence in the alignment tool and the native guide. This is because the solved native structure does not have those 9 residues fully defined, the N-terminus is disordered. You can see this in the above video clip of the native structure.

Sometimes the functional parts of proteins are disordered until they meet their native partner, such as zinc fingers (which need the zinc to fold).
This will happen during CASP; they might give us a long sequence to work on, but once the structure has been solved they will only be able to assess a subset of that sequence in our predictions.

We will try to do our best during CASP to trim regions that are obviously disordered, until then we hope this Quest to the Native will help you improve your alignment tool skills for CASP 9!

If you want to examine this native structure in more detail:
http://www.rcsb.org/pdb/explore/images.do?structureId=2KGT

steveB Lv 1

If you weight the voting results according to global soloist ranking how do the percentages come out ?

beta_helix Staff Lv 1

we should look into that!

Also, this is just one data point… we need to do this for more puzzles in order for any results to be statistically significant!