openmm
Petition to support parameterization using `CGenFF` for small molecules
Hello,
First of all, thank you tremendously for developing the openmmforcefields tool, especially for allowing the on-the-fly parameterization using GAFF and others! It really saved a lot of pain of dealing with file format compatibility!
In terms of small molecule parameterization, besides the currently available options including GAFF, is there a plan to add support for CGenFF? With the current support for CHARMM force field, it would be really amazing to have CGenFF in the toolkit too :)
I saw a similar question in the FAQ section. I am not certain if there is a conda-installable tool for generating CGenFF parameters. Meanwhile, the CGenFF developers have built a tool and web server to streamline the parameterization process: https://docs.silcsbio.com/2023/cgenff/cgenff.html#usage and https://cgenff.silcsbio.com/. And it is also in CGenFF's developers' interest (and their future plan) to make their force field more widely used: https://cgenff.silcsbio.com/commonFiles/futureProspects.php
Integrating CGenFF into this package would almost surely require a command-line tool or API that provides the result of force field parameterization. (Think along the lines of tleap and friends within AmberTools or the OpenFF Toolkit for SMIRNOFF force fields.) I haven't used CGenFF (or the CHARMM ecosystem) much does such a tool exist? Bonus points for a standalone tool not a part of a larger or monolithic software suite.
I haven't used a SilcsBio product myself - how is CGenFF licensed? Your second link is from 2018 and looks more like an academic website and less like the other pages with SilcsBio branding.
After discussion, we have decided that an openmmforcefields interface to CGenFF (by means of a new template generator akin to what is already available for GAFF) will remain unsupported, unless the way that it is being distributed changes in the future. Obtaining a license to use CGenFF as a standalone program (rather than the online interface through which it is currently made available) is difficult. Also, since no source is available, only binaries, through the licensing process, we can't make guarantees about the stability of its input/output formats, which could make maintenance more difficult.
