Cysteine crash

Case number:845829-2006149
Opened by:Aubade01
Type:DevPrev Issue
Opened on:Sunday, October 28, 2018 - 22:38
Last modified:Thursday, November 1, 2018 - 04:55

Added info:

Working on cysteine segments 41 and 67. Wiggling a zero length band eventually crashed client.

It was observed that a band between the white cysteine HYDROGEN and its opposite cysteine will not form the disulfide bridge no matter how close and can cause the client crash. A band carefully placed on the SULFUR of each cysteine will form the disulfide bridge in a normal way.

>>>>> Aubade00/01

(Sun, 10/28/2018 - 22:38  |  3 comments)

LociOiling's picture
User offline. Last seen 2 hours 17 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

I see two other feedbacks on the same topic. I'm consolidating the comments here. Not 100% sure, but the comments all seem to refer to the "Disulfide/Histidine" puzzle in devprev.

DEVPREV - Disulfide/Histidine

Crashing on manual rebuild around cysteine. Also In View menu when View Sidechains changed, Cysteine sidechains banded but not bonded -- disappear!
Unable to bond cysteines even when very close. (Windows 10)

>>>>> Aubade 00/01


Bonded cysteines remain visible, but unbonded ones don't when Don't Show (Fast) selected.

Client crash remains and difficulty in bonding when rebuilding around closely paired cysteines.

>>>>> Aubade 00/01

LociOiling's picture
User offline. Last seen 2 hours 17 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

A few thoughts on this issue.

Something about cysteines has changed, and it seems to have happened with the big update this summer. The recipe Bridge Wiggle is no longer very effective in forming new bridges. Manually pulling and flipping sidechains seems to be more effective.

Since the big update, we've had several toxin revisiting puzzles with multiple disulfide bridges. While old techniques don't work very well, it's still possible to get all the bridges formed as expected. I don't recall having many crashes along the way.

Maybe there's a problem with the "Disulfide/Histidine" puzzle in particular, since I think it's been around a while.

I haven't tried banding the hydrogens at the tips of the cysteines. Those hydrogens go away when a disulfide bridge forms, reducing the atom count of each cysteine by one. I wouldn't be too surprised if that's causing a crash.

Disulfide bridges are revealed when "show bonds (sidechain)" is on, and should remain visible unless they get broken.

If you've banded a sidechain, it should remain visible, regardless of the "view sidechains" option selected. One exception is in design puzzles, where mutating a banded sidechain deletes the band, allowing the sidechain to disappear if "don't show" is checked.

I'd expect cysteine to get hidden if "don't show" is checked and there's no disulfide bridge (or if "show bonds (sidechain)" is not checked).

brow42's picture
User offline. Last seen 1 year 36 weeks ago. Offline
Joined: 09/19/2011
Groups: None

ooooo, yes. Having bands connected to atoms that disappear is a bad thing. Especially if connected to the last atom (atom 11 in a reduced cysteine) Usually there's another atom with the same index after the change, and the band connects to that. I should take those H-H bands out of bridge wiggle if you choose to wiggle.

This is also a problem with mutation, which is why I remove bands before mutating anything.

I never actually use the wiggle function of bridge wiggle, since it didn't produce the gains I was hoping for.

It's pretty much unfixable, except for jflat to drill down and validate and remove the bands after every rosetta call.


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, Boehringer Ingelheim, RosettaCommons