Nbfix for OPLS-AA

Hi all,

I may have asked this before, so apologies in advance. We keep bumping into the issues with Na+ and Ca2+ in OPLS-AA and I was wondering if someone already implemented nbfix for this forcefield in Gromacs. There is literature that suggests nbfix parameters, but we are sort of hoping someone already implemented and tested it. Of course, if no one has done it, we will have to face the music…
We stick with OPLS-AA, because all the nanomaterial models we use were designed for OPLS-AA down to the mixing rules, etc. In fact, I just transferred one of the models into CHARMM and while the ion-related issues are gone, we are now expectedly facing problems related to the nanomaterial (i am guessing mostly because of the fudge factor of 1.0 in CHARMM), so for us it’s either working out nbfix in OPLS, or going completely nuts trying to fix the nanomaterials in something like CHARMM.

All suggestions are welcome!

Since there are no responses, here is a “smaller” question, and once again, I think it was discussed here. Is it possible to specify a fudge factor that is not equal to the ff default for particular 1-4 pairs without writing out funny charges or new modified LJ parameters?
Example: we have a sheet of hexagonal boron nitride (hBN) and the atoms are massively charged (+0.9 and -0.9 for boron and nitrogen, respectively). Under OPLS-AA with its fudge of 0.5, the whole thing is perfectly stable. When the parameters are copy-pasted into CHARMM (generally a highly questionable idea, I know), it distorts. However, when I rescale all the Coulomb charges by 1/sqrt(2) (no change to LJ), everything is back to normal for the hBN sheet. This to me is a good indication that the distortion is mainly due to CHARMM’s default 1-4 factor of 1.0 applied to Coulomb.
I would like to emulate the majority of what kept the hBN’s geometry in the original ff – it is perfectly sufficient for our purposes. Hence, is it possible to cleanly specify a fudge factor of 0.5 for B-B, N-N, and B-N pairs so that it will selectively override the default of 1.0?

Also, our situation is a bit different from that described here. We have edge atoms, which will have charges different from the +/-0.9 mentioned above, while the atomtypes are the same as for the ones with charges +/-0.9. What we want is that the fudge factor is 0.5 for whatever the charges are, as listed in the atoms section of the topology.


Let’s narrow down the question even more. Beyond changing the function type from 1 to 2 in the molecular topology for everything under [pairs] and setting custom interactions under [pairtypes] in ffnonbonded with five parameters, is anything else needed? Thank you.