Lipid14 - dihedral multi-periodicity problem again

rafwiewiora commented 8 years ago

@swails

Same problem as in ff14ipq with breaking convention for multi-periodicity dihedrals in the .dat, slightly different flavor here - in lipid14.dat all dihedrals with the Lipid14 v2.10 (paramfit) comment are listed like this:

cB-cB-cD-cD   1    0.3464        0.000           1.000     SCEE=1.2 SCNB=2.0 Lipid14  v2.10 (paramfit)
cB-cB-cD-cD   1   -0.5577        0.000           2.000     SCEE=1.2 SCNB=2.0 Lipid14  v2.10 (paramfit)
cB-cB-cD-cD   1   -0.2920        0.000           3.000     SCEE=1.2 SCNB=2.0 Lipid14  v2.10 (paramfit)
cB-cB-cD-cD   1   -0.0943        0.000           4.000     SCEE=1.2 SCNB=2.0 Lipid14  v2.10 (paramfit)
cB-cB-cD-cD   1    0.0226        0.000           5.000     SCEE=1.2 SCNB=2.0 Lipid14  v2.10 (paramfit)

For now only had a system that used two of these problematic dihedrals, for those LeAP had all periodicities. They don't intertwine here as they did in 14ipq so it might be that LeAP gets this 'right' for all of them.

Do you think ParmEd should further relax the parsing?

rafwiewiora commented 8 years ago

Btw I am not converting lipid11 for now (ever perhaps if there's lipid14), because it's using mixed scaling factors and we have no support.

rafwiewiora commented 8 years ago

@swails and reminder about this too

swails commented 8 years ago

ParmEd and tleap behave differently here? This parameter file violates the defined Amber format. I'll post a bug report for this, as well. Thanks!

swails commented 8 years ago

Bug report filed.

rafwiewiora commented 8 years ago

ParmEd and tleap behave differently here?

Yeah, ParmEd doesn't expect another periodicity if it didn't have a negative number before right? Thanks!

rafwiewiora commented 8 years ago

Closing this for now - no conversion for lipid14 until problem is resolved on the AMBER side.

choderalab / openmm

Lipid14 - dihedral multi-periodicity problem again #11