pdb2gmx generates wrong dihedrals for disulfide bridges using gromos force-fields (all of them)
The bug is present at least in gmx2020.x and 2021.x (bonds and angles are correctly generated)
To reproduce it: gmx_mpi pdb2gmx -f 2d4f.pdb -ignh
disulfide is between 277 and 878 in this case
these are the offending lines (in topol.top [ dihedrals ]) :
273 275 276 277 1 gd_34
273 275 278 280 1 gd_42
273 275 278 280 1 gd_45
109 108 110 124 1 <- wrong
108 110 111 112 1 gd_34 <- wrong
275 278 280 282 1 gd_14
278 280 282 296 1 gd_43
...
874 876 877 878 1 gd_34
874 876 879 881 1 gd_42
874 876 879 881 1 gd_45
323 326 330 332 1 <- wrong
876 879 881 883 1 gd_14
879 881 883 896 1 gd_43
879 881 883 896 1 gd_44
881 883 884 885 1 gd_34
Edited by Andrey Alekseenko