Hi,
In the Gromacs port of the Charmm forcefield, Charmm36m-jul2021.ff, ffbonded.itp the following dihedrals are given:
CN3 CN3 NN2B CT2 9 180.000000 46.024000 2 ; UPN
NN3A CN4 NN2 CT2 9 180.000000 46.024000 2 ; APN
These are provided for peptide nucleic acids. The first relates to uracil, and the second to adenine. I’m presuming uracil is intended to be used for thymine as well, because there is no entry marked TPN in ffbonded.itp but three of the four terms marked UPN are dihedrals in TPN. Note that the entries for the peptide nucleic acids refer only to the dihedrals that span between the backbones and the bases, as all the base-only dihedrals are just the same as in e.g. DNA.
The .rtp entries for the thymine and adenine peptide nucleic acids are given below:
[ TPN ]
;
[ atoms ]
C C 0.5100 1
O1’ O -0.5100 1
C2’ CT2 -0.0300 2
H2’ HB2 0.0900 2
H2’’ HB2 0.0900 2
N2’ N -0.3300 2
C5’ CT2P 0.0000 2
H5’ HB2 0.0900 2
H5’’ HB2 0.0900 2
N NH1 -0.4700 3
H1’ H 0.3100 3
C6’ CT2 -0.0200 3
H6’ HB2 0.0900 3
H6’’ HB2 0.0900 3
C3’ CC 0.4300 4
O3’ O -0.5400 4
C4’ CT2 -0.0700 4
H4’ HA2 0.0900 4
H4’’ HA2 0.0900 4
N1 NN2B -0.3400 5
C6 CN3 0.1700 5
H6 HN3 0.1700 5
C2 CN1T 0.5100 5
O2 ON1 -0.4100 5
N3 NN2U -0.4600 5
H3 HN2 0.3600 5
C4 CN1 0.5000 5
O4 ON1 -0.4500 5
C5 CN3T -0.1500 5
C5M CN9 -0.1100 5
H51 HN9 0.0700 5
H52 HN9 0.0700 5
H53 HN9 0.0700 5
[ bonds ]
+N C
C O1’
C C2’
C2’ H2’
C2’ H2’’
C2’ N2’
N2’ C5’
C5’ H5’
C5’ H5’’
C6’ C5’
C6’ H6’
C6’ H6’’
C6’ N
N H1’
N2’ C3’
C3’ O3’
C3’ C4’
C4’ H4’
C4’ H4’’
C4’ N1
N1 C2
N1 C6
C2 O2
C2 N3
N3 H3
N3 C4
C4 O4
C4 C5
C5 C5M
C5 C6
C6 H6
C5M H51
C5M H52
C5M H53
[ impropers ]
C2 N1 N3 O2
C4 N3 C5 O4
C5 C4 C6 C5M
N -C C6’ H1’
C O1’ C2’ +N
C3’ O3’ C4’ N2’
N2’ C2’ C5’ C3’
[ APN ]
;
[ atoms ]
C C 0.5100 1
O1’ O -0.5100 1
C2’ CT2 -0.0300 2
H2’ HB2 0.0900 2
H2’’ HB2 0.0900 2
N2’ N -0.3300 2
C5’ CT2P 0.0000 2
H5’ HB2 0.0900 2
H5’’ HB2 0.0900 2
N NH1 -0.4700 3
H1’ H 0.3100 3
C6’ CT2 -0.0200 3
H6’ HB2 0.0900 3
H6’’ HB2 0.0900 3
C3’ CC 0.4300 4
O3’ O -0.5400 4
C4’ CT2 -0.0700 4
H4’ HA2 0.0900 4
H4’’ HA2 0.0900 4
N9 NN2 -0.0500 5
C5 CN5 0.2800 5
N7 NN4 -0.7100 5
C8 CN4 0.3400 5
H8 HN3 0.1200 5
N1 NN3A -0.7400 5
C2 CN4 0.5000 5
H2 HN3 0.1300 5
N3 NN3A -0.7500 5
C4 CN5 0.4300 5
C6 CN2 0.4600 5
N6 NN1 -0.7700 5
H61 HN1 0.3800 5
H62 HN1 0.3800 5
[ bonds ]
+N C
C O1’
C C2’
C2’ H2’
C2’ H2’’
C2’ N2’
N2’ C5’
C5’ H5’
C5’ H5’’
C6’ C5’
C6’ H6’
C6’ H6’’
C6’ N
N H1’
N2’ C3’
C3’ O3’
C3’ C4’
C4’ H4’
C4’ H4’’
C4’ N9
N9 C4
N9 C8
C4 N3
C4 C5
N3 C2
C2 N1
N1 C6
C6 N6
N6 H61
N6 H62
C6 C5
C5 N7
N7 C8
C8 H8
C2 H2
[ impropers ]
N6 C6 H61 H62
C6 N1 C5 N6
N -C C6’ H1’
C O1’ C2’ +N
C3’ O3’ C4’ N2’
N2’ C2’ C5’ C3’
From the .rtp entries we then have the two following dihedrals, if we follow the bonds:
TPN: C4’ (CT2) is bonded to N1 (NN2B) is bonded to C6 (CN3) is bonded to C5 (CN3T) hence CT2 NN2B CN3 CN3T
APN: C4’ (CT2) is bonded to N9 (NN2) is bonded to C4 (CN5) is bonded to N3 (NN3A) hence CT2 NN2 CN5 NN3A
In UPN the dihedral CN3 CN3 NN2B CT2 is present, which is intended, but the equivalent sequence of atoms for TPN: CN3T CN3 NN2B CT2 is not present in ffbonded.itp
In APN the dihedral listed in ffbonded.itp (NN3A CN4 NN2 CT2) is not present in the .rtp; NN2 is adjacent to a CN4 but this CN4 is not adjacent to NN3A. On the other hand, NN2 is adjacent to CN5 which is adjacent to NN3A and consequently the dihedral NN3A CN5 NN2 CT2 is present in the structure but not present in ffbonded.itp
Am I right then to assume the following:
NN3A CN4 NN2 CT2 listed in ffbonded.itp as ‘APN’ actually refers to NN3A CN5 NN2 CT2
CN3T CN3 NN2B CT2 needs to be added to ffbonded.itp
The URL for the download of the forcefield is: MacKerell Lab
I did a quick test of topology generation with the original parameters, new parameters, and with the lines deleted. Topologies generated, and simulations ran, in all three cases. I used gmx energy to evaluate proper dihedral energy terms of the starting structures and found a very small difference in the initial values when the new terms were used (+0.8kJ/mol) and no difference between the original values and when the lines were simply deleted but this was not a rigorous test.
I’ve been using Charmm36m for over a year now but am concerned I might be misunderstanding something so would appreciate a second set of eyes on this. Namely - what dihedral is being used when the lines are deleted? Grompp is not giving any errors in all three cases (line deleted, lines changed, and original).
- Jack