CHARMM-36 FF glycoprotein simulation: error no residue type find on rtp

GROMACS version: 2020.6
GROMACS modification: No
Dear GMX users,
I’d tried to simulate the SARS-COV-2 Spike glycoprotein obtained from CHAMM-GUI into GROMACS using charmm36-feb2021.ff.tgz (obtained from: MacKerell Lab) but on pdb2gmx I faced this error:
[… many lines of information …]
Opening force field file /opt/gromacs-2020.6/share/gromacs/top/charmm36-feb2021.ff/merged.arn

Program: gmx pdb2gmx, version 2020.6
Source file: src/gromacs/gmxpreprocess/pdb2gmx.cpp (line 745)

Fatal error:
Atom N in residue BGLC 0 was not found in rtp entry BGLC with 24 atoms
while sorting atoms.
For more information and tips for troubleshooting, please check the GROMACS
website at Errors - Gromacs

CHARMM36 defines some building blocks (BGLC, BGLCA, BGLCA0, BGLCN0, BGLCNA) on merged.rtp file and since pdb2gmx reads 3-long residue type characters (columns 18-20) and CHARMM36 presented 5 or 6-long residue name on merged.rtp file, so how can I correct assign these N-glycosilated residues on my protein generating my top files?

CHARMM-GUI provides everything you need. You should not try to generate a topology within GROMACS. It can’t handle such complex glycoproteins. CHARMM-GUI also has a repository of prebuilt models for these systems.

Many thanks for you quick response Dr Justin!
I saw the COVID-19 Proteins Library at CHARMM-GUI site but I’m interested to simulate its soluble form (Fully-glycosylated full-length S protein complex models) available only on PDB format. I don’t want to run this membrane-embebbed protein, yet.
I also tried to adapt the CHARMM-GUI GROMACS files (TOP, GRO) deleting the membrane, counter-ions,… but it required many text-file editions and it halted after 15 steps of position restrained steepest descent (curiously, the same step for membrane-embebbed protein!!).
New tips will be a lot helpful!

You should be able to build whatever you like using CHARMM-GUI starting from whatever protein coordinate file, since it can do complex things like add fly cans and handle the non-linear topology. My point is that pdb2gmx essentially can’t be used in this case (at least not in any easy way) so you need to generate all your inputs using CHARMM-GUI and at no point should you be trying to produce a topology using GROMACS tools.