Handling SYBYL atom type in Mol2 parser

[ppxasjsm]

When loading a molecule form amber prm7 and rst7 and then writing it out to mol2, the atom types are set to dummy, this means the mol2 file cannot be read by other mol2 file readers.

The code below should reproduce this.

import BioSimSpace as BSS
ethane = BSS.IO.readMolecules(['ethane.prm7','ethane.rst7'])
BSS.IO.saveMolecules('ethane_mol2', ethane,"mol2")

mol2_issue.zip

[lohedges]

When writing the Mol2 parser we decided not to attempt to infer the SYBYL atom type. This means that Mol2 files written by Sire following a format conversion can't be read by other Mol2 parsers.

If anyone has a comprehensive reference for this format, or a link to existing code that robustly implements a way to infer the type, then let me know. (I might be able to use RDKit to get the type, that add that as a molecule property before writing with BioSimSpace.)

[lohedges]

I had previously looked at the specification at the end of this document, although it is quite technical to implement this in code.

Someone is currently working on a Mol2 writer for RDKit here. Once this is merged, we could use this to generate the SYBYL atom type for any molecules in the system. (They can infer the SYBYL atom type from their internal molecule representation.)

I've also looked at other packages (OpenBabel and ParmEd) and they don't infer the type on format conversion either.

[chryswoods]

Agree - better to reuse rdkit code once they get it working. It is a very challenging problem which has many ambiguous edge cases.