Rama map broken on 1311

Case number:845813-2003154
Topic:Game: Tools
Opened by:Susume
Status:Open
Type:Bug
Opened on:Friday, December 2, 2016 - 02:08
Last modified:Monday, December 5, 2016 - 19:22

Rama map not displaying correct dot position. Shared solution "1311 Rama map wrong on 10 serine". First pic shows rama position of dot for 10 serine. Grab that dot and drag a tiny bit in any direction, and that arm of the protein rotates almost 180 degrees, while rama dot barely moves - see 2nd pic. It appears the first rama position is wrong for the actual angles of the residue, and using the rama map to set new angles produces the actual angles that make sense for the area of the rama map where the dot appears.

AttachmentSize
foldit_1480643808.png359.96 KB
foldit_1480643821.png374.18 KB
(Fri, 12/02/2016 - 02:08  |  5 comments)


Susume's picture
User offline. Last seen 6 days 11 hours ago. Offline
Joined: 10/02/2011

Same problem on different puzzle (1312). Similar fold, same error on same residue. Not sure if Remix put the same shape on that loop or if I somehow managed to do it by hand. Shared solution "1312 Rama map wrong on 10 serine". 1st pic shows 10 serine's dot; drag it a tiny bit and that arm of the protein shifts almost 180 degrees while rama dot barely moves (2nd pic).

Joined: 04/24/2014
Groups: None

Thanks Susume, I've let the team know. :)

bkoep's picture
User offline. Last seen 10 hours 16 min ago. Offline
Joined: 11/15/2012
Groups: None

Hi Susume, it's hard to tell from the photos, but it looks like the peptide bond between residues 10 and 11 is originally in a cis (as opposed to trans) conformation. I think Rama Map adjustments, by default, will automatically set the torsion of adjacent peptide bonds to perfect trans conformation—in this case, rotated exactly 180 degrees.

As you may know, cis peptide bonds are very rare in natural proteins (with certain exceptions, of course), and 99% of the time they indicate a problem with the protein model. However, this is clearly unreasonable behavior in the Rama Map, and should be fixed. Thank you for bringing it to our attention!

P.S. I'm curious about the origins of the cis peptide bond in your model—do you know how it came about (e.g. manually closing a cutpoint, Rebuild, Remix)?

Susume's picture
User offline. Last seen 6 days 11 hours ago. Offline
Joined: 10/02/2011

When I turn on "show backbone issues" in view menu there is no exclamation point on this loop.

I looked back through my saves on 1312 and found the following: I created the 4-residue sheet-sheet turn at 9-12 (DSPR) using the rama map (no cutpoint) as a red-red-red-green "ideal" loop. It was stable in that shape until I ran the DRemixW script on it, using length 6 remixes only, on residues 8-13. The script remixes, shakes and wiggles (no rebuilds). That script changed it to the shape you see in the pics, red-blue-blue-blue, with the bad rama dot on the serine. So there is a remix in the database that creates this problem.

I have shared the position right before I ran the DRemixW script as "1312 before DRemixW". When I remix 8-13 by hand on this pose, I get 2 remixes, neither of which is red-blue-blue-blue but both of which have problems. First remix: 9-12 are blue-blue-blue-red with unideal backbone warning (not a CIS warning) between 9 and 10; 9 has a bad rama dot (but 10 rama dot is ok). 2nd remix: 9-12 are blue-yellow-blue-blue, no backbone warning, but 10 has a bad rama dot (9 rama dot is ok). It is possible that one of these wiggled out to the position I originally complained about, or that after other loops in the protein had been remixed by the script, remix found a further change to make to this area.

The history of 1311 is the same - I created the turn using the rama map (no cuts), most likely as red-red-red-green, and subsequently ran DRemixW on it. Other than wiggle by hand at different clash importance, that is the last thing I did to it.

bkoep's picture
User offline. Last seen 10 hours 16 min ago. Offline
Joined: 11/15/2012
Groups: None

Ah, I see. Residue 11 is a proline, which is unusual in that it is equally happy with either a cis or trans connection to residue 10. So, contradicting my comment above, the cis peptide bond at SER 10 was perfectly reasonable before the Rama Map converted it to trans. The Rama Map is most definitely at fault here.

Sometimes Remix will return loops with cis peptide bonds. In the database these are common next to a PRO residue, but Remix doesn't check the sequence of your selection so these cis bonds can appear next to any residue type. When you manually Remix residues 8-13 in the case above, the first Remix result yields a cis peptide bond at residues 9-10, which is problematic for ASP 9 and SER 10, and triggers an unideal backbone warning; the second Remix result yields a cis peptide bond at residues 10-11, which is allowed by PRO 11.

Sitemap

Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, RosettaCommons