Improved PDBWriter performance using range & caching (#4472)

* Improvements to PDBWriter's performance

Author: BFedder

package/CHANGELOG

@@ -15,7 +15,7 @@ The rules for this file:
 
 -------------------------------------------------------------------------------
 ??/??/?? IAlibay, HeetVekariya, marinegor, lilyminium, RMeli,
-         ljwoods2, aditya292002, pstaerk, PicoCentauri
+         ljwoods2, aditya292002, pstaerk, PicoCentauri, BFedder
 
  * 2.8.0
 
@@ -32,6 +32,7 @@ Fixes
  * Fix deploy action to use the correct version of the pypi upload action.
 
 Enhancements
+ * Improved performance of PDBWriter (Issue #2785, PR #4472)
  * Added parsing of arbitrary columns of the LAMMPS dump parser. (Issue #3504)
  * Documented the r0 attribute in the `Contacts` class and added the 
    `n_initial_contacts` attribute, with documentation. (Issue #2604, PR #4415)

package/MDAnalysis/coordinates/PDB.py

@@ -149,6 +149,7 @@
 import logging
 import collections
 import numpy as np
+import functools
 
 from ..lib import util
 from ..lib.util import store_init_arguments
@@ -1028,6 +1029,7 @@ def _write_next_frame(self, ts=None, **kwargs):
         self._check_pdb_coordinates()
         self._write_timestep(ts, **kwargs)
 
+    @functools.cache
     def _deduce_PDB_atom_name(self, atomname, resname):
         """Deduce how the atom name should be aligned.
 
@@ -1217,7 +1219,7 @@ def validate_chainids(chainids, default):
         else:
             atom_ids = np.arange(len(atoms)) + 1
 
-        for i, atom in enumerate(atoms):
+        for i in range(len(atoms)):
             vals = {}
             vals['serial'] = util.ltruncate_int(atom_ids[i], 5)  # check for overflow here?
             vals['name'] = self._deduce_PDB_atom_name(atomnames[i], resnames[i])