search

choderalab / openmm

OpenMM is a toolkit for molecular simulation using high performance GPU code.
1 stars 0 forks source link

ffAM1 / ffPM3 - LEaP overrides nonbonded equivalencing #7

Closed rafwiewiora closed 8 years ago

rafwiewiora commented 8 years ago

@swails

ffAM1 fails energy validation on nonbonded.

parmAM1.dat specifies equivalencing, but then specifies different params for types that should be equivalenced. 

C   C*  CA  CB  CC  CN  CM  CK  CQ  CW  CV  CR  CA  CX  CY  CD
(...)
  CA          1.9061  0.0860000
  CM          1.9108  0.0860000
  C           1.9127  0.0860000

ParmEd overrides different params with equivalencing, LEaP appears to act in the opposite way - frcmod from prmtop for villin:

NONB
N3    1.8197   0.1700
H     0.6099   0.0157
CT    1.9344   0.1094
HP    1.0911   0.0157
HC    1.4568   0.0157
C     1.9127   0.0860
O     1.6306   0.2100
N     1.8126   0.1700
H1    1.3616   0.0157
OH    1.7156   0.2104
HO    0.0000   0.0000
O2    1.6555   0.2100
CA    1.9061   0.0860
HA    1.4564   0.0150
S     1.9922   0.2500
N2    1.8126   0.1700
C*    1.9127   0.0860
CW    1.9127   0.0860
H4    1.4078   0.0150
NA    1.8126   0.1700
CN    1.9127   0.0860
CB    1.9127   0.0860
CC    1.9127   0.0860
NB    1.8126   0.1700
CR    1.9127   0.0860
H5    1.3554   0.0150

CA != C:

CA    1.9061   0.0860
C     1.9127   0.0860
rafwiewiora commented 8 years ago

Same thing for ffPM3 / parmPM3.dat

jchodera commented 8 years ago

D'oh.

swails commented 8 years ago

Ugh! Sounds like we need to fix the ParmEd parser and add a warning.

swails commented 8 years ago

Should be fixed in ParmEd/ParmEd#601 -- give it a try if you want.