openforcefield / openff-benchmark

Comparison benchmarks between public force fields and Open Force Field Initiative force fields
MIT License
10 stars 2 forks source link

Swope analysis fails when running over a dataset containing molecules with multiple 3 letter company codes #101

Closed ldamore closed 2 years ago

ldamore commented 3 years ago

From the csv files it can be seen that molecules from different companies are wrongly being compared.

qm_min,mm_conf,mm_min,rmsd (mm_conf/qm_min),dE (mm_conf-mm_min) JNS-01356-07,BRI-01356-07,BRI-01356-08,, 6.84774323e+00 JNS-01356-07,BRI-01356-02,BRI-01356-08,, 9.72582731e+00 JNS-01356-07,BRI-01356-04,BRI-01356-08,, 1.41296403e+00 JNS-01356-07,BRI-01356-03,BRI-01356-08,, 3.09150186e+00 JNS-01356-07,BRI-01356-01,BRI-01356-08,, 1.25368682e+01 JNS-01356-07,BRI-01356-05,BRI-01356-08,, 7.29534843e+00 JNS-01356-07,BRI-01356-00,BRI-01356-08,, 3.39467115e+00 JNS-01356-07,BRI-01356-06,BRI-01356-08,, 8.77350580e+00

Note: not tested yet, however lucas analysis may exhibit the same wrong behaviour.

dotsdl commented 2 years ago

Closed by #102.