Good evening Everyone,
sorry if I am re-opening this thread so late, but in the end we preferred sharing the (possible) bug here. Specifically, it is a matter of using the free energy code when adopting the mdp file keywords:
couple-moltype = 'choose_solute_to_decouple'
couple-lambda0 = vdw-q
couple-lambda1 = none
and corresponding non-null coul_lambdas series, the whole of this with a charged molecule to be decoupled from the environment.
I understand the large advantage of this approach with respect to having to prepare a double topology .top file with A and B states, which is why we were using it when willing to solely compute the solvation energy. But you will see no warning regarding net charge is returned in none of the 3 examples prepared for you (please see attached link): in each one you can find the straightforward script executing the ‘grompp’ and ‘mdrun’ commands.
To face the problem in the meanwhile, I ask whether one could add for instance an ion of the box to the solute (the one to be decoupled) topology, so to turn both off contemporarily; thus the ion turning into a free dummy atom with null intermolecular interactions and surrounded by the solvent. Or if this is completely unreasonable, and one should instead do the proper morphing of an ion to a water (or vice-versa), having to go back to the double topology picture (A and B states with all the corresponding parameters in one top file).
no_error_if_annihilating_charged_solute
Thank you in advance for all your work and explanations.
Bests,
Jacopo