crash on small molecule design test

Case number:954892-2004494
Topic:Developer Preview
Opened by:LociOiling
Status:Open
Type:Bug
Opened on:Sunday, November 26, 2017 - 22:11
Last modified:Wednesday, November 29, 2017 - 02:41

In the current small molecule design test puzzle, Foldit crashed with an unhandled exception when I tried to add a carbon to replace a hydrogen.

I don't see a way to attach a text file to a feedback, so here's an excerpt from log.txt:

core.chemical.icoor_support: Resetting the ICOORD invalidated some CHI entries - recomputing.
Reprotonate: 
core.chemical.icoor_support: Resetting the ICOORD invalidated some CHI entries - recomputing.
Smiles: CC(O)C(C)(N)O
ENERGY: 7449.1
Fixing hydrogens.
core.chemical.icoor_support: Resetting the ICOORD invalidated some CHI entries - recomputing.
core.chemical.icoor_support: Resetting the ICOORD invalidated some CHI entries - recomputing.
internal_coords_atom:
autodeterimine_chi:
FinalE finalize:
replacing residue
setting game pose 
marking chemical change 
NonMatching vertex: 2 :Name:  C1  :ResIndex 199
delta_score: -58.2645
Playing sound: 3
Generated 15 rotamers for LG_5005 in 0.094s total.
rotate mouse down

UNHANDLED EXCEPTION
  1: library_main +4318205 bytes (no line)
  2: library_main +3793258 bytes (no line)
  3: library_main +3791871 bytes (no line)
  4: library_main +5349259 bytes (no line)
  5: library_main +5353612 bytes (no line)
  6: library_main +5361162 bytes (no line)
  7: library_main +5354121 bytes (no line)
  8: library_main +3140622 bytes (no line)
  9: library_main +3143687 bytes (no line)
 10: library_main +3565532 bytes (no line)
 11: library_main +3357082 bytes (no line)
 12: library_main +5112115 bytes (no line)
 13: library_main +357976 bytes (no line)
 14: library_main +3356012 bytes (no line)
 15: library_main +5117173 bytes (no line)
 16: library_main +5106611 bytes (no line)
 17: SetManipulationInputTarget +83 bytes (no line)
 18: DispatchMessageW +2298 bytes (no line)
 19: CallWindowProcW +151 bytes (no line)
 20: glDebugEntry +3763 bytes (no line)
 21: SetManipulationInputTarget +83 bytes (no line)
 22: DispatchMessageW +2298 bytes (no line)
 23: CallWindowProcW +151 bytes (no line)
 24: GetTeamViewerInterface +18252 bytes (no line)
 25: SetManipulationInputTarget +83 bytes (no line)
 26: DispatchMessageW +2298 bytes (no line)
 27: DispatchMessageW +596 bytes (no line)
 28: DispatchMessageA +16 bytes (no line)
 29: library_main +5115338 bytes (no line)
 30: library_main +3105955 bytes (no line)
 31: library_main +16713 bytes (no line)
 32: library_main +21 bytes (no line)
 33: no symbol (no line)
 34: no symbol (no line)
 35: no symbol (no line)
 36: BaseThreadInitThunk +36 bytes (no line)
 37: RtlSubscribeWnfStateChangeNotification +1081 bytes (no line)
 38: RtlSubscribeWnfStateChangeNotification +1028 bytes (no line)
(Sun, 11/26/2017 - 22:11  |  5 comments)


LociOiling's picture
User offline. Last seen 7 hours 37 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

Here's the complete log.txt, not sure why I didn't have the option on the original post.

LociOiling's picture
User offline. Last seen 7 hours 37 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

Another crash happens when I select the ligand and click on the Ligand Queue (Q) tool. Unlike the previous crash, this one is highly reproducible.

LociOiling's picture
User offline. Last seen 7 hours 37 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

Just piling on this feedback, not sure if this is a related problem.

I attempted to build an AZT molecule just to see what happened. Keep in mind that I don't have much of a clue about what I'm doing.

In JMol, the AZT ligand shows three Ns double-bonded to each other. I was able to add the first N, but received a large synthesizeability penalty (-5000 range) for the second. Oddly, I was able to eventually clear the penalty.

For me, the reasons for the penalty appearing and disappearing are unclear.

At some point, a debug.txt is generated, stating "Encountered molecule which cannot properly be represented in RDKit." This has happened at least twice. In one case, the client eventually crashed, something like 20 minutes later. In the second case, no crash.

I think the error may have occurred when adjusting bonds (single to double or double to single), but it's hard to be sure.

I've attached log.txt and debug.txt from one instance, but of course it's not clear if debug.txt is related to the crash.

I've shared AZT0009 with scientists, the final version that's at least close to the JMol model.

Also shared AZT003a and AZT004b, intermediate steps, and AZT0006, which has the big penalty.

LociOiling's picture
User offline. Last seen 7 hours 37 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

On reflection, the synthesizeablility penalty may relate to pH. I'll see if keeping the pH more or less balanced while building the molecule helps.

LociOiling's picture
User offline. Last seen 7 hours 37 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

I've been able to reproduce the "RDKit" error, where a debug.txt is generated, but no crash occurs. The solution AZT003a can be used as a base.

Solution AZT003a - AZT starter molecule: Changing bonds may generate a debug.txtSolution AZT003a - AZT starter molecule: Changing bonds may generate a debug.txt

The steps to reproduce this error are as follows.

1. Open solution AZT003a, shared with scientists.

2. Select the ligand.

3. Select the ligand design tool (L).

4. Select atoms 6 and 14 as shown in the image.

5. Click the double bond button.

The debug.txt is generated at this point, even though the molecule has not changed.

Alternately, the following steps do not produce the error:

4a. Select atoms 5 and 6.

5a. Click the single bond button.

6. Select atoms 6 and 14.

7. Click the double bond button.

No debug.txt is generated in this case, and the molecule is updated as desired.

If you create this molecule on your own, I suspect the atom numbers will be different.

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, Microsoft, Adobe, RosettaCommons