Torsion filtering doesn't seem to work

[claggner]

Describe the bug

Hi, I'm running unidock2 with use_tor_lib: true set in the unidock_conf.yaml file, but the output files still show a lot of structures where amide and carboxylic acid groups are connected to an aromatic ring at a ~90 degree torsion angle angle. Shouldn't the torsion library help reduce the occurrence of such structures?

Expected behavior

Less strained ligand geometries, especially for groups that are in conjugation with aromatic systems (amides, acids, esters, nitro groups, ...)

To Reproduce

No response

Environment

No response

Additional Context

No response

[Hong-Rui]

Hi, can you provide your input protein and ligand files as well as the docking center and YAML cofigurations, so that we can do some test on this system. Thanks.

[claggner]

Thank you for your reply, @Hong-Rui. My current structures are proprietary, but I'll try to put something together that I can share.

[claggner]

OK, here's an example run. I ran unidock2 -r 7lck_rec_prot.pdb -l chembdrugs_filtered_3D_prot.sdf -c 126.86 159.71 116.85 -cf unidock_conf.yaml > unidock_torsion_example.log 2>&1. Then I processed the output file in KNIME to get the top pose for the 500 best-scoring molecules by vina_binding_free_energy. Look for example at the geometries for molecules 7, 12, 13, 14, 40, 99, 159.

torsion_example.zip