I've got about 60 similar, small molecule inhibitors with calculated and biological values (energies, charges, dipoles, HBD/HBA, etc. and IC50's respectively). We would like to know what features of these molecules make them good vs bad inhibitors. We did some basic regression, but none of the biological values could easily be attributed to a particular calculated value (all R^2 < 0.85). I'd like to go back and cast a finer net to look at more specific properties (dihedrals, distances between functional groups, etc). The structures are about 80% similar, and I'd like to automate the process of getting this additional information from each structure. Problem is that no two structures have the same numbering for their atoms (carbon 1 or benzene ring 1 is not the same in each molecule), so scripting something quick and dirty to get the values won't work since I wouldn't be comparing the same sets of atoms from each structure. I've tried converting back and forth between a few formats (SMILES, SDF, etc) to see if this would at least get the similar pieces on the same numbering system, but this didn't work. Is there an easy way to do this or will I have to do this all by hand? Thanks!
