smina with flexible molecules

[Gonzalo Colmenarejo]

Hi,

thank you for your excellent software and all your kind help.

Is there a way to limit the time smina spends in one molecule? I'm having issues with a ligand set that apparently contains some very flexible molecules and the time it is taking in each of these is much more than one minute, while for more rigid structures it is about 10 sec. If I could limit the time taken at each molecule I'd avoid this delay. Using multi_smina.py I had initially estimated (based on the 10 sec / mol and 40 cores) to screen ca. 200K molecules in about 14 hours, but now it is taking more than four days. What's the best way for dealing with these flexible molecules that can appear in your set of ligands to avoid them clogging your calculations?

Thanks a lot

Gonzalo

[David Koes]

The reason larger molecules take longer is the number of steps to perform in each Monte Carlo chain is scaled heuristically with the size of the molecule and number of rotatable bonds. There is no way to change this in smina, but in gnina (https://github.com/gnina/gnina) there is a num_mc_steps commandline option that sets the number of steps to a fixed amount. This does not put a cap on the heuristic amount, but takes a fixed number of steps so smaller molecules may be oversampled.

Implementing a max_mc_steps option that caps the heuristic is a good idea.