[WIP] implement PDB fixes for #52

[IAlibay]

This is WIP.

Opening as there are things which need to be documented as I go along.

TODO:

• Remove ANISOU records
• Remove icodes
• Switch caps to HETATM
• Rename NMA to NME
• Fix hydrogen naming for caps + first residue
• Separate out non-protein non-solvent non-ion entries from protein.pdb to individual sdf files
  • bace (TLA)
  • cmet (7PE)
  • eg5 (ADP)
  • pfkfb3 (POP)
  • shp2 (PO4)
• Make sure SSBOND entries are consistent
• Remove altloc entries

Notes:

• Thrombin is a minefield of a PDB file when it comes to icodes.
  a. There are many icode entries throughout the various chains
  b. There are non-contiguous residue numbers, see GLU H 217 being followed by GLY H 219
  c. I'm not sure there should be an ACE cap on chain I (to be chased up in Fix protein.pdb files; remove GROMACS-specific files; drop LFS usage #52)
• Caps and post-ACE first residues had alternate namings. Naming is now in line with primary atom names from: https://github.com/openmm/openmm/blob/dd80fb3fdcba6a75803f8e68f444a657fe42d4f5/wrappers/python/openmm/app/data/pdbNames.xml

[jchodera]

Thanks so much for tackling this, @IAlibay! Copying @bobym to monitor.

Just to check, @IAlibay : Are you having to correct all of these files manually? If so, we should have @bobym report these issues to be fixed in Maestro's export or protein prep wizard.

[IAlibay]

@jchodera - yeah so currently most of these fixes are by-hand because I don't want to disrupt the HELIX / SHEET / TURN entries. The exception here is for Thrombin where manually fixing the icode insertions for those entries + the ATOM/HETATM entries was too much.

That being said, if we don't care about the HELIX/SHEET/TURN metadata, I would advocate for re-doing all these things in one go with MDAnalysis' parser (except from the SSBOND entry stuff which has to be manual right now), and could even add in the missing bonds for the caps. This would make things a lot more reproducible.

With regards to reporting back upstream: I remember @jbluck trying this a couple of months back - he might have some insights from his own discussions with the Schrodinger folks on the matter.

[dotsdl]

Closing as no longer relevant.