Newer versions of ParmEd errantly converting to CHAMBER-style prmtop #1151
Description
I have been using ParmEd in Python to make some minor changes to Amber prmtop files:
parmed_top = parmed.load_file(path_to_prmtop)
<do some actions on parmed_top>
parmed_top.save(path_to_prmtop, overwrite=True)
I originally began development on version 3.2.0+27, and had no trouble. Since updating to version 3.4.0, however, the same code causes ParmEd to output a CHAMBER-style file. To illustrate, the portion of my prmtop file before the %FLAG ATOM_NAME line goes from
%VERSION VERSION_STAMP = V0001.000 DATE = 02/26/21 10:37:27
%FLAG TITLE
%FORMAT(20a4)
default_name
%FLAG POINTERS
%FORMAT(10I8)
68801 23 49734 19187 8064 5150 16509 15763 0 0
147758 15841 19187 5150 15763 97 213 214 57 1
0 0 0 0 0 0 0 1 56 0
15398
... to:
%VERSION VERSION_STAMP = V0001.000 DATE = 02/27/21 15:21:13
%FLAG CTITLE
%FORMAT(20a4)
%FLAG POINTERS
%FORMAT(10I8)
68801 57 49734 19187 8064 5150 16509 15763 0 0
147758 15841 19187 5150 15763 97 213 214 1 0
0 0 0 0 0 0 0 1 56 0
15398
%FLAG FORCE_FIELD_TYPE
%FORMAT(i2,a78)
1 CHARMM force field: No FF information parsed...
... (along with extensive changes to the rest of the file, of course) whereas, on version 3.2.0+27, the prmtop file produced by ParmEd is identical to the input file (except for the date and timestamp, of course). On both versions, the chamber AmberParm attribute is False, as expected.
This bug is causing problems for my workflow (namely, passing the topology to OpenMM, which cannot read CHAMBER-style files). For now, I'll just revert to an earlier version without the bug.
Thanks!