luancarvalhomartins
Error "The largest distance between excluded atoms is X.XXX nm, which is larger than the cut-off distance." using GROMACS 2022
This is an effect of how (upstream bug 4051) is being handled. However, this also means that there is a slightly problem of excluded atoms interacting in the intermediate $\lambda$ windows. I don't think it would affect the FEP prediction, ad the end points should be correct (because the charges and VdW of the non-interacting state are set by direct modification of the topology). Nevertheless, this could reduce the overlap between $\lambda$ windows and causes crashed when using GROMACS 2022.
It would be possible to replace the exclusions for a combination of [pairs_nb] and [nonbond_params], as mentioned in the upstream bug ticket. I will further think about that possibility.
For now, please, use GROMACS < 2022 if you are affected by this issue.
With the release of GROMACS 2023, the 2021 versions are no longer maintained. I tried to run with GROMACS 2023 and 2022, and both releases still presents the issue you explain in this ticket, pushing us to use a non-maintained version of GROMACS if we want to use PyAutoFEP. I am not sure how problematic this can be, but I thought the best was to comment that version 2023 does not solve the issue. Great work with PyAutoFEP!
Thank you very much for your message. Agreed. This should be high priority now. It will require some quite tricky compensation of charges, as explained in the upstream bug report. Hope I can work on it soon.
Note that for small, localized perturbations this bug will not manifest (as the exclusion distance will be smaller than the cutoff. Nevertheless, it's a deal-breaker priority.