You are here

I made a RDKit Chem.Mol to params converter — feedback welcome!

2 posts / 0 new
Last post
I made a RDKit Chem.Mol to params converter — feedback welcome!

For a project I was really in need of a clean way to convert a RDKit Chem.Mol object to a params file as the way I was making params files was horrid.

So I ended up writing a MIT-licenced python module that can load and save params files and do "surgery" on them or create one from a RDKit Mol object, which can include the dummy atom (*) which becomes a connect entry. It can be used without either rdkit installed (surgery only!) or pyrosetta installed (no testing method).

It was fairly straightforward, so I won't be devastated if someone tells me that you can load a mol file directly in pyrosetta... (I seriously suspect I am doing it wrong given the PDB components library and the chemical submodule in pyrosetta)  —is there?

I had to make several guesses with the AtomType allocation, but I think they are mostly right —in the repo's example folder there is a params file generated as a ligand. I did not copy from the file because I didn't really want to get into licencing issues. I was actually unaware that there were two forms (aa and generic), so I wrote generic first, found it did not work, and re-did it with the aa form — is there a tag for a params file with generic atom types or a different way to add them to a pose? 

I have not released it to pip yet and there are some tweaks needed and I need to do some minor changes (as listed in the repo).

But I thought I best ask in case I get some good suggestions. Thanks!

Post Situation: 
Sun, 2020-05-03 10:14

Hey Matt,


This is pretty incredible!!!  I really wish we had a nice, public repository for python modules that you could add this to!  We are working on it, and its been something on my list for quite some time, and hopefully we will have something soon.

At this point, I'd like to formally invite you and your lab to join into the RosettaCommons.  You may be the first to join in this way, but you've coded some incredibly useful things, started to be an integral part of the community, and I think it would be pretty awesome to have you join in.  If you are interested, shoot me an email, and I'll work toward getting this going. 

- Jared Adolf-Bryfogle

Institute for Protein Innovation

Protein Design Laboratory

Sun, 2020-05-03 19:38