GROMACS version: any above and including 2016
GROMACS modification: No
Dear GROMACS users,
I am looking for the person who ports the CHARMM36 FF to GROMACS. One of my colleagues needed a special amino acid (deprotonated Arg) which is part of CHARMM but (understandably) did not make it to the GROMACS port. As it also needs special atom types, a simple RTP entry is not sufficient.
I started to write a CHARMM to GROMACS porting program (András Wacha / charmm2gmx · GitLab), and while trying to validate it against the official CHARMM36 port downloaded from Alexander MacKerell’s page, I found some minor discrepancies which I would like to ask about.
nucleic acids are linked in the other direction (in CHARMM, a bond between +P and O3’ is defined, while in your port it is between -O3’ and P; not that it matters much, but still…)
Residue SAM: O3’ is of atom type ON5, while in your port it has ON2 type.
The C-O bond is missing in your CO residue topology entry
in the recent version of your CHARMM port (from February, 2021) the atom naming in ACE and NME have changed, probably to reflect the upstream naming (e.g. to CAY from CA, to CY from C in ACE and to NT from N in NME). This, however, breaks the "-C, N, CA, C, +N) cmap interaction, defined in the topology of the next residue. CHARMM explicitly lists the CMAP interactions in the ACE and CT3 patches.
I have written an e-mail to your Virginia Tech address.
I can confirm the ACE changes in the latest port. , I had problem with it when I had Ace protected peptide. I did not wanted tinker with the cmap interactions, so I changed the atom names back to the older or original names for ACE in the merged.rtp and it worked.
This is the first I write in this forum. Excuse me if I would not use it properly. You said that you have conversion scripts charmm to gromacs?
The problem is the following, and I am wondering that any of these scripts would solve it:
Using Charmm, I produce psf, crd and pdb files of my protein, which is a dimer (proa and prob, identical sequences). The psf and pdb files contain the accurate mentions ‘PROA’ and ‘PROB’ at the expected lines. Using the ‘charmm2gmx_prot.str’ as written in Charmm-gui, I successfully produce the itp files required for Gromacs. The problem is charmm2gmx_prot.str produces a monomer instead of the dimer. Not any atoms are lost, and the residues are numbered as if it was a protein twice as long as the original monomers. Obviously ‘prob’ numbering starts at n+1, with n the number of residues in proa. I am sure that it will run correctly on Gromacs. However, to simplify the analysis afterwards, I would prefer to simulate a ‘proa.itp’ and a ‘prob’itp’, each one starting at, say residue 1, instead of a single long entity.
What would you suggest?