search

drugdata / D3R

Drug Design Data Resource is a suite of software to enable filtering, docking, and scoring of new sequences from wwpdb.
Other
22 stars 10 forks source link

Evaluation of target 5mva on week 48 in 2017 ran slowly #170

Open coleslaw481 opened 6 years ago

coleslaw481 commented 6 years ago

Evaluations took a long time (30,000+ seconds per evaluation) for 2017 week 48. Digging into the final.log files for the various evaluations, it was seen that 5mva took a long time to evaluate. Here is example output from final.log from stage.7.33567_autodockvina.extsubmission.evaluation

11/29/17 02:17:55: =================We start to work on this target 5mva=================
11/29/17 02:18:05:      Successfully create the crystal object
11/29/17 04:12:41:      Successfully calculate the distance between original LMCSS ligand vs crstal ligand. Distance is 19.3610851452
11/29/17 06:08:14:      Successfully calculate the rmsd for this category hiTanimoto, the rmsd is 19.3988457341
11/29/17 08:02:31:      Successfully calculate the rmsd for this category SMCSS, the rmsd is 19.2488521247
11/29/17 09:58:50:      Successfully calculate the rmsd for this category LMCSS, the rmsd is 19.1135471324
11/29/17 11:54:20:      Successfully calculate the rmsd for this category hiResHolo, the rmsd is 19.1086867012
11/29/17 11:54:20: ++++++++++++++++Finished for this target 5mva+++++++++++++++++

The above logging is from d3r/evaluate.py and the logging is not adequate to isolate which schrodinger command is being slow. Suggest one adds more logging by default

j-wags commented 6 years ago

5MVA has 23 chains with ADP as a ligand. ADP has at least 6 self-symmetries. Evaluation of 5MVA took 6-7 hours. The longest other evaluation took 5 minutes. I don't think that 6*23 combinations explains this factor of difference, especially because some other ligands should have self-symmetries as well. I'll need to come back and figure this one out.