Weird side chains in Puzzle 352

Case number:845799-988123
Topic:Biochem
Opened by:infjamc
Status:Closed
Type:Bug
Opened on:Monday, August 2, 2010 - 03:19
Last modified:Monday, August 22, 2011 - 21:54

Is it only me, or are the prolines in these puzzle actually supposed to be hydroxy-prolines? (This bug is showing up only on this puzzle.)

==> If they are... it seems to me that the post-translational modification patch files aren't compatible with Foldit yet. (Try changing the side chain configuration and you'll see what I mean.)

AttachmentSize
foldit_1280718774.png250.03 KB
foldit_1280719088.png310.6 KB
(Mon, 08/02/2010 - 03:19  |  5 comments)


infjamc's picture
User offline. Last seen 16 hours 20 min ago. Offline
Joined: 02/20/2009
Groups: Contenders

***Update***

My impression is that the side chain would take a huge penalty (try -3000 or worse) upon changing its configuration either because the ring would not close properly or because the hydroxyl group is not being connected correctly. The issue was already noted when the post-translational modification patch file was created back then: there are two possible locations where a hydrogen on the C(4) atom of proline could be added, and the program could guess incorrectly.

Suggestion: Manually disable the patch file in Foldit for now if the prolines are not supposed to be HYP, and manually fix the rotamer samples of hydroxyprolines if they are intended to be there...

infjamc's picture
User offline. Last seen 16 hours 20 min ago. Offline
Joined: 02/20/2009
Groups: Contenders

Oops, there was a typo in my last post:

"Two possible locations where a hydrogen on the C(4) atom of proline could be added" should read "Two possible locations where a hydrogen on the C(4) atom of proline could be replaced by the added hydroxyl group."

beta_helix's picture
User offline. Last seen 1 hour 48 min ago. Offline
Joined: 05/09/2008
Groups: None
Status: Open » Open

Thanks for pointing this out infjamc,

Looking at the CASP9 webpage, all the Prolines in this CASP target are supposed to be regular Prolines:

http://predictioncenter.org/casp9/target.cgi?id=190&view=refinement

None of them are supposed to be hydroxy-prolines, but it seem that Rosetta changed this during the idealization process (to fix the bug you had noticed here: http://fold.it/portal/node/988083) and put a hydroxyl group onto the C-gamma atoms!

We will look into manually disabling the patch file in Foldit.

Until then, try to treat them as regular Prolines (easier said than done, of course!) thanks for catching this!

infjamc's picture
User offline. Last seen 16 hours 20 min ago. Offline
Joined: 02/20/2009
Groups: Contenders

Thanks for the quick response. I must say that I was quite surprised when the prolines showed up as hydroxyprolines, as my having participated in the development of those patch files last year gave me the impression that Rosetta--and, by extension, Foldit--does not have the capability to process post-translational modifications by default. Still, I wonder why the idealization process would put hydroxyl groups on there if the only things that are being tweaked are bond lengths and bond angles. After all, the patch is only supposed to be applied when there is an oxygen atom bonded to C(4).

==> Whatever the cause may be, we might have just discovered a bug related to the patch files that was undiscovered back then. Perhaps the Rosetta users should be warned of this issue as well? Then again, we now know at least two things:

1. The bad news: The patch files can be loaded in Foldit when they're not supposed to (which is a bit embarrassing considering that I happen to be the original author of the patch file).

2. The good news: Surprisingly, the patch files can already be loaded in the current implementation of Foldit. When can we design an enzyme with on/off switches via phosphorylation? Before 2020? ;-)

Joined: 06/17/2010
Status: Open » Closed

Old puzzle. Closing.

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