timemachine icon indicating copy to clipboard operation
timemachine copied to clipboard

Published 20 hours ago • verified publisher icon proteneer

Align Mols by Core

Open badisa opened this issue 2 years ago • 1 comments

An issue was seen in finding the MCS of two compounds due to slight conformational changes. This PR adds the ability to set up a set of restraints between two compounds in vacuum to align their cores together. This can be used to construct a valid MCS mapping that is stable for SingleTopologyV3

Test Case Two cyclohexanes that have different conformations. Screenshot from 2022-08-31 14-07-44

If mapped using mcs_map and get_core_by_mcs the hydrogens mapped are invalid due to the relative position to the ring Screenshot from 2022-08-31 14-10-02 When this core is applied to SingleTopologyV3, the initial conformation is wrong and is an unstable transformation. Screenshot from 2022-08-31 14-08-16

Aligning to the carbon ring then running mcs_map and get_core_by_mcs, the mapping is as expected. Screenshot from 2022-08-31 14-10-14

badisa avatar Aug 31 '22 20:08 badisa

Been running this on CDK2, Hif2a, SHP2 from here: https://github.com/MCompChem/fep-benchmark/

So far things are looking pretty good. Takes about 15 seconds per pair at the moment (32 core machine, but looks to only be about 2 cores really doing work)

badisa avatar Sep 01 '22 20:09 badisa