PDB-XYZ Conversion

This directory contains my Python scripts for converting from PDB to XYZ and from XYZ to PDB. These scripts are a workaround for TINKER's pdbxyz and xyzpdb, which can fail with nucleic acids and solvated systems with numerous WAT/HOH residues.

Notes:

pdbxyz-for-amber.py works for AMBER parameter sets.

pdbxyz-for-amoeba.py works for AMOEBA bio, pro, and nuc parameter sets.

• This assumes a generic HIS is HIE (protonated at N-epsilon). If you only use HIS in your PDB, make sure you modify different protonation states to HIP (both N-epsilon and N-delta) or HID (N-delta) naming.

xyzpdb.py works for both AMBER and AMOEBA parameter sets.

• Residue writing is limited to 9999 total residues, before starting back at 1. This doesn't seem to be an issue for visualization with VMD, but may be important for other programs.

pdbxyz4amber-pmd-params.py is meant to work with prmtop-based parameter files created using generate_TINKER_parameters.py. This is because the .prm files the script generates are written in a standardized format, where atom name strings are composed of only the residue's 3-letter code and atom name. The bulk of the original Python scripts were processing the parameter files distributed with the TINKER program. xyzpdb4amber-pmd-params.py will do the conversion back to a PDB using the same prmtop-based parameter files.

A Note on AMBER Parameters

• The TINKER distribution of amber96.prm, amber99.prm, amber99sb.prm uses Glutamic Acid to refer to the AMBER GLU NOT GLH. As such, GLH is not included in the parameter sets, and will need to beadded as a "non-standard." amber94.prm and amber98.prm both use Glutamic Acid (COOH) for GLH.

How

The first few lines specify the infile, outfile, and prmfile. Change these to reflect your system.

These scripts work by taking the atom lines from an input parameter file and reading them into a pandas dataframe. The residue name from the name string (ex: "Isoleucine CB") is converted to its 3-letter code (or some general 3-letter code), and that field is split into multiple columns (ResName and AtomName).

Important: Name strings cannot have more than 1 space, as this will throw an error in parsing the parameter file.

For best results, give any non-standard residues unique atom names, and use the wanted 3-letter code in their name string.

In pdbxyz, the residue names and atom names are matched from this table, and the atom.mass is set to the TINKER atom type. In xyzpdb, the TINKER atom type is used to set the residue and atom names.

xyzpdb for AMOEBA has the added challenge of protein residue types defaulting to alanine for the backbone (excluding proline and glycine). To address this, the insertion_code attribute is set to the residue number. Later on, the insertion_code is removed so that the final PDB is numbered correctly.

The reverse of this issue is that some non-standard residues fall back on known atom types. In that case, then you'll likey want to add some if statements after the comment that says Catch non-standard problems here! in pdbxyz.

Incomplete Conversion

If pdbxyz cannot find an atom type, the residue/atom it had issues with will be printed to the Terminal. Additionally, the final XYZ will have ATOM TYPE NOT FOUND following the connectivity.

If xyzpdb cannot find a residue name, the residue name will be assigned as FIX. A warning that this occurred will be printed to the terminal.

If you're using xyzpdb for a visualization program, check that each segment has the requisite TER cards.

Errors Encountered with Analyze

Sometimes when using AMBER parameter files, TINKER analyze will say there are missing angles or torsions. These parameters are likely to be defined in the relevant AMBER parm files. Add them to your TINKER parameter file and you should be golden.

Future plans:

• Consolidate for-amber and for-amoeba programs.
• Add argparse / something for command-line input.
• Investigate limits to residue numbering/ways to get around that.
• Fix TER cards after protein in xyzpdb.
• Maybe actually formulate this into a Python package with tests?
• Fix DX5 HO5' atoms classed incorrectly with AMBER prm files (1244 instead of 1245).