Query, ENH: supporting lambda scaling for MARTINI 3 lipid diversity
One of the most important steps in an MD simulation workflow is the setup of the initial configuration--for massive lipid-based vesicle/virus systems we often face the challenge of an unknown lipid composition, and unknown ratio between inner and outer leaflet populations. As a result of this fact combined with the sizes of these systems, steric conflicts in the process of system construction are hard to avoid.
GROMACS has been an essential tool in relaxing these steric conflicts prior to energy minimization. The lambda scaling/free energy "abuse" mechanism of this steric conflict resolution is well known (and published, i.e., https://doi.org/10.1021/ct501111d) and even suggested by GROMACS proper when energy minimization fails in some cases I believe.
I've created a small (2 molecule) reproducer repo that works with GROMACS 2024.2 (https://github.com/tylerjereddy/gmx_martini3_alchembed_repro) for a problem that happens with some MARTINI 3 lipid types. It would be great if there were a workaround for this. In this particular case, it seems to be the forcefield/parameters for MARTINI 3 POPI molecule. Steps to reproduce with that repo cloned locally:
gmx grompp -f alchembed-cg.mdp -c system.gro -p topology.top -o alchembed.tpr -maxwarn 20gmx mdrun -v -s alchembed.tpr -deffnm alchembed_1x_lipids -nt 1 -ntmpi 1
Should fail with:
Program: gmx mdrun, version 2024.2
Source file: src/gromacs/topology/topsort.cpp (line 142)
Fatal error:
Function type Restr. Angles does not currently support being perturbed in free
energy calculations