Alternative to CGenFF for generating large ligand topology

[scamander]

I am now preparing a ligand topology following GROMACS protein-ligand tutorial.
However, I have a large ligand (446 atoms).

$ wc -l my_ligand.pdb
446 my_ligand.pdb

And

$ head my_ligand.fix.mol2
@<TRIPOS>MOLECULE
my_ligand
 717 726 0 0 0
SMALL
GASTEIGER

@<TRIPOS>ATOM
      1 N          20.2110  -14.1580    4.4710 N.3     1  144_ligand       -0.3199
      2 H          20.9960  -14.3850    5.0530 H       1  144_ligand        0.1189
      3 CA         19.8650  -12.7390    4.4190 C.3     1  144_ligand        0.0830

When I tried to upload my_ligand_fix.mol2 to CGenFF server, I noticed it the maximum number of atoms allowable is 384.

Note that my_ligand_fix.mol2 has gone through processing right up to sort_mol2_bonds.pl output.

Is there a way to get around this?
Or what other alternative to CGenFF that I can use instead?

-Sca

[jalemkul]

One should never attempt to parametrize something that is hundreds of atoms all at once, using any method. Break the compound into smaller, representative model compounds, and develop parameters for those. What ligand has 446 atoms? Based on the names and atom types, it appears to be a peptide; in that case, you do not need CGenFF and you can produce the entire topology with pdb2gmx.

[scamander]

Hi Justin,

Thank you for the reply.
It is indeed a peptide!

What command and steps should I use to make that peptide topology with pdb2gmx ?
Thanks and truly hope to hear from you again.

Is there any official GROMACS tutorial for peptide-protein binding complexes?

-Sca

[jalemkul]

The two protein chains should be separated by TER or given distinct chain identifiers. There is no special command to issue; pdb2gmx understands separate chains if they are properly notated in the input PDB file.

[scinikhil]

gmx pdb2gmx understand different chains and process it.