GROMACS version: 2021.3
GROMACS modification: Patched with PLUMED
Hi everyone,
When building a processed topology (topology file with all #include statements replaced by actual content) with the grompp -pp option, I noticed differences in energies between the original and processed topologies for simulation reruns. I am using the Amber99SB_ILDN force field. When digging a little deeper, I found out that the following section was missing in the processed topology file:
#define torsion_ILE_N_CA_CB_CG2_mult1 0.0 0.8158800 1 ; Proteins 78, 1950 (2010)
#define torsion_ILE_N_CA_CB_CG2_mult2 0.0 -3.5396640 2 ; Proteins 78, 1950 (2010)#define torsion_LEU_C_CA_CB_CG_mult1 0.0 2.3890640 1 ; Proteins 78, 1950 (2010)
#define torsion_LEU_C_CA_CB_CG_mult2 0.0 -1.4978720 2 ; Proteins 78, 1950 (2010)
#define torsion_LEU_C_CA_CB_CG_mult3 0.0 0.5648400 3 ; Proteins 78, 1950 (2010)#define torsion_ASP_N_CA_CB_CG_mult1 0.0 -11.024840 1 ; Proteins 78, 1950 (2010)
#define torsion_ASP_N_CA_CB_CG_mult2 0.0 -4.978960 2 ; Proteins 78, 1950 (2010)
#define torsion_ASP_N_CA_CB_CG_mult3 0.0 -0.029288 3 ; Proteins 78, 1950 (2010)
#define torsion_ASP_N_CA_CB_CG_mult4 0.0 1.769832 4 ; Proteins 78, 1950 (2010)
#define torsion_ASP_N_CA_CB_CG_mult5 0.0 0.970688 5 ; Proteins 78, 1950 (2010)
#define torsion_ASP_N_CA_CB_CG_mult6 0.0 -0.891192 6 ; Proteins 78, 1950 (2010)#define torsion_ASP_CA_CB_CG_OD_mult1 0.0 0.0 1 ; Proteins 78, 1950 (2010)
#define torsion_ASP_CA_CB_CG_OD_mult2 0.0 -1.853512 2 ; Proteins 78, 1950 (2010)
#define torsion_ASP_CA_CB_CG_OD_mult3 0.0 0.0 3 ; Proteins 78, 1950 (2010)
#define torsion_ASP_CA_CB_CG_OD_mult4 0.0 -0.577392 4 ; Proteins 78, 1950 (2010)
#define torsion_ASP_CA_CB_CG_OD_mult5 0.0 0.0 5 ; Proteins 78, 1950 (2010)
#define torsion_ASP_CA_CB_CG_OD_mult6 0.0 -0.054392 6 ; Proteins 78, 1950 (2010)#define torsion_ASN_C_CA_CB_CG_mult1 0.0 2.389064 1 ; Proteins 78, 1950 (2010)
#define torsion_ASN_C_CA_CB_CG_mult2 0.0 -2.493664 2 ; Proteins 78, 1950 (2010)
#define torsion_ASN_C_CA_CB_CG_mult3 0.0 0.493712 3 ; Proteins 78, 1950 (2010)
#define torsion_ASN_C_CA_CB_CG_mult4 0.0 -1.744728 4 ; Proteins 78, 1950 (2010)
#define torsion_ASN_C_CA_CB_CG_mult5 0.0 0.435136 5 ; Proteins 78, 1950 (2010)
#define torsion_ASN_C_CA_CB_CG_mult6 0.0 -0.422584 6 ; Proteins 78, 1950 (2010)#define torsion_ASN_CA_CB_CG_ND2_mult1 0.0 -4.376464 1 ; Proteins 78, 1950 (2010)
#define torsion_ASN_CA_CB_CG_ND2_mult2 0.0 -0.757304 2 ; Proteins 78, 1950 (2010)
#define torsion_ASN_CA_CB_CG_ND2_mult3 0.0 -0.146440 3 ; Proteins 78, 1950 (2010)
#define torsion_ASN_CA_CB_CG_ND2_mult4 0.0 0.418400 4 ; Proteins 78, 1950 (2010)
#define torsion_ASN_CA_CB_CG_ND2_mult5 0.0 0.543920 5 ; Proteins 78, 1950 (2010)
#define torsion_ASN_CA_CB_CG_ND2_mult6 0.0 -0.443504 6 ; Proteins 78, 1950 (2010)
This section is present in the bonded parameters. Command used to generate the processed topology:
$ gmx grompp -p topol.top -pp processed.top -c solv_ions.gro -f min1.mdp
grompp seems to be skipping this section when building the processed topology. Admittedly I was a bit lazy and only tested the 2019 version, so I do not know if this is fixed in later versions. Hopefully this can help others experiencing a similar issue.
Kind regards,
Olivier