No default improper dihedral types in Amber ff14sb-OL21 for DNA simulations

User discussions
, ,
1

GROMACS version: 2022.4
GROMACS modification: No

Hello,

I’m trying to set up a DNA simulation with Amber OL21, using the Gromacs port of the ff14sb-OL21 forcefield. But after I created the topology I got a bunch of the same error for different improper dihedrals when I try to run gmx grompp:

ERROR 1 [file topol.top, line 705]:
  No default Improper Dih. types

But the weird part is that when I checked some of the dihedrals they’re defined on the ffbonded.itp file of the forcefield. and others even if you try to defined them they don’t work

For example atoms 9 11 12 23. (CT - N* - CM - C).
This dihedral exists in the ffbonded:

C CM N* CT 4 180.00 4.18400 2

But It still gives the same error, I have tried even copying and pasting this dihedral with the same atom order as it’s on the topology, but it doesn’t fix the errors.

If I try adding an improper dihedral that doesn’t seem to exist in the ffbonded.itp file, it simply doesn’t work even though the formatting is correct. I have done it multiple times with other forcefields.

If I were to change the function type 4 (which is supposed to be used for improper dihedrals) to 2 it reduces the number of errors but creates a new error

ERROR 1 [file ffbonded.itp, line 847]:
  Too many parameters or not enough parameters for topology B

I’m really on a dead end right now, I’m only using the amber forcefield because it was supposed to be better for DNA.

All my pertinent files can be found in this link:

MD-FIles for DNA simulation with amber

Any help will be appreciated,

Thank you

2

How did you generate the topology? I can see that your force field files use dihedral type 4 (as typical for Amber) but your topology file has 2 (more common for the CHARMM family). This depends on the [ bondedtypes ] entry in the .rtp files that were used to generate the topology.

So, what happens is your topology asks for type 2 dihedral CT - N* - CM - C and your parameters have type 4 dihedral CT - N* - CM - C, and these are not the same. They have different functional forms and different numbers of parameters, which is why you can’t just change the interaction type and expect it to work.

There’s a great explanation of all that in table 14 of the manual, hope that makes it clear.

3

Thank you for your reply, I generated the topology with the gmx pdb2gmx module.
I ended up switching to amber OL15 as it works fine, but I might try again.
It made it quite clear. Thanks again!