Anomalies in the beginning of RDFs

GROMACS version:2021.4
GROMACS modification: No

I’ve ran several simulations with mixtures of methyl-THF (forcefield & parameters from charmm36) and water. For some reason, when running gmx rdf with following command line, I get weird artifacts in the beginning of RDF graphs. I checked the boxes and they haven’t phase separated.

gmx_mpi rdf -f md-297.xtc -s md-297.tpr -o thfm-thfm_rdf.xvg -ref THFM -sel THFM

Box size is 200 molecules.
Simulation parameters:

Could this be something to do with the simulation settings or box size?

Posting the whole view of RDF as a reply.

Those are all intramolecular distances. You need to adjust nrexcl to exclude all intramolecular nonbonded interactions and regenerate a .tpr file for use with gmx rdf. That topology shouldn’t be used for anything else as it is not physically sensible.