Need help minimising a backmapped system (I used backward method)

GROMACS version:2023.3
GROMACS modification: Yes/No
Hi everyone,

I’m using backward.py and/or the initram script for backmapping. During the energy minimization step, I get errors when non-bonded interactions are turned off. It looks like atoms are overlapping in the initial configuration, which prevents proper minimization.
Has anyone dealt with this issue before? Any advice on how to handle overlapping atoms or best practices for minimizing complex backmapped membranes would be appreciated.

My EM.mdp file looks like this

;define=-DFLEXIBLE
integrator=steep
nsteps=20000
emstep=0.1
pbc=xyz
cutoff-scheme = Verlet

; Table extension is needed initially
table-extension=2

; During first steps nonbonded interactions
; are excluded within groups membrane and protein
;energygrps=Protein Membrane Solvent
;energygrp_excl=Protein Protein Membrane Membrane

; Usually no trajectory is written,
; but this can be changed (-trj)
nstxout=0

Potential Energy = 4.9176702e+17
Maximum force = inf on atom 334
Norm of force = inf

Thanks,