Closed j-wags closed 3 years ago
The charge/protonation states on the NTerm GLU SMIRKS are indeed explicit, so I don't think there could be confusion with the charged AA being a substructure of the protonated AA
<Proper
smirks="[H][C@@]([C]=O)(C([H])([H])[C:2]([H])([H:1])[C:3](=[O:4])[O-])[N+]([H])([H])[H]"
periodicity1="2"
phase1="0.0 * degree"
id="A14SB-NTerminal-GLU-HC_2C_CO_O2"
k1="0.0 * mole**-1 * kilocalorie"
idivf1="1.0"></Proper>
<Improper
smirks="[H][C@@]([C]=O)(C([H])([H])[C:1]([H])([H])[C:2](=[O:4])[O-:3])[N+]([H])([H])[H]"
periodicity1="1"
phase1="180.0 * degree"
k1="3.5 * mole**-1 * kilocalorie"
id="A14SB-NTerminal-GLU-2C_O2_CO_O2"
idivf1="1.0">
</Improper>
If I rerun with only GLU
, the I get the exact same values for NTerm GLU. So this isn't a case of parameters for another AA being incorrectly substituted into the GLU sidechain
This improper center is very bent, by virtue of its proximity to the positively charged N terminus.
Depending on how I measure around the improper center, I can get angle values ranging from 150.8 and 152.5 degrees.
k*(1+cos(periodicity*theta-phase))
periodicity1="1" phase1="180.0 * degree" k1="3.5 * mole**-1 * kilocalorie"
import math
k=3.5
per=1.
ph=180.
theta = 152.5
k*(1+math.cos(math.radians(per*(theta-ph)))) * 4.184
27.63338664106188
theta=150.8 k*(1+math.cos(math.radians(per*(theta-ph)))) * 4.184
27.427070899539093
It looks like, at some point in the past 40 days, we fixed this. Current output reads:
NTerminal/GLU/GLU
HarmonicBondForce 2.1379218101501465 kJ/mol
HarmonicAngleForce 6.882795810699463 kJ/mol
PeriodicTorsionForce 46.8884162902832 kJ/mol
NonbondedForce -415.76202392578125 kJ/mol
-359.8529052734375 kJ/mol
[H][C@@](C(=O)N([H])C([H])([H])[H])(C([H])([H])C([H])([H])C(=O)[O-])[N+]([H])([H])[H]
HarmonicAngleForce 6.882845401763916 kJ/mol
HarmonicBondForce 2.1379215717315674 kJ/mol
NonbondedForce -415.7628173828125 kJ/mol
PeriodicTorsionForce 46.889320373535156 kJ/mol
-359.85272216796875 kJ/mol
67 amber dihedrals (3 impropers) and 67 off dihedrals
GLU
NTerm GLU has a large torsion energy discrepancy (~20 kJ/mol). CTerm ASP is off by 3 kJ/mol. Both discrepancies are in torsions.
Top suspects:
Next step would be to unpack the parameters from each system and see what's being mis-assigned
Related values for reference:
CTerm isn't that bad:
For comparison, here's mainchain GLU/GLH:
ASP
Mainchain ASP/N/H
CTerm ASN/P
Nterm