Acqueous anion force field parameterization

[Matteo Peluso]

Hello,

I am trying to parameterize the non bonded force field of a bromide ion in water exploiting the ForceBalance approach.

I have performed Gaussian calculation of 32 and 128 water molecules cluster and built the qdata.txt and all.gro in order to apply a fitting.

I have done a few attempts but the results weren't fully satisfying. I am then writing here to ask if I 'missed' something during the fitting procedure, I am attaching the input file.

PS: I wrote a parser that from a set of Gaussian output files build the qdata.txt and all.gro, if you think it would be useful in the future I'll push it on Github.

Best Regards

Matteo

[Lee-Ping Wang]

Hello Matteo,

It looks like you are fitting vdW parameters of bromide ion using ab initio calculations of ions in water clusters as the fitting data. This may be difficult because your ab initio calculation includes the energies of all the water molecules and their interactions with each other, which your fitting cannot account for. I recommend that you use the interaction energy target to do this, or use the abinitio target where the conformations of the water molecules are not changing across your conformations and only fit the energies as the position of the ion changes (not the forces).

Thanks,

- Lee-Ping

[Matteo Peluso]

Hello Lee-Ping,

Yes I am trying to fit the vdW parameters of bromide ion.

As you said the energy of the entire cluster is not a proper descriptor for this parameterization, is there a way to consider a complexation energy? Or the energy of just the ion?

Would it be possible to consider for the fitting just the forces on the ion itself and not of the entire cluster?

Thanks for the reply and for the advices to exploit at his best the FB procedure

[Lee-Ping Wang]

Hello Matteo,

> As you said the energy of the entire cluster is not a proper descriptor for this parameterization, is there a way to consider a complexation energy?

Yes, you can use the Interaction or Binding target types to do this. The Binding target is designed for flexibility to accommodate various different systems (e.g. including water dimer and hexamer in the same target), whereas the Interaction target is designed to accommodate a single system at different structures.

Please check the studies/005_iamoeba example calculation on how to do that. I haven't used the binding energy target with Gromacs but I hope it should work automatically. Let me know if something unexpected happens.

> Or the energy of just the ion?

I'm afraid there's no straightforward way to do that.

> Would it be possible to consider for the fitting just the forces on the ion itself and not of the entire cluster?

Yes, you can fit the forces on just the first n atoms in the AbInitio target using "fitatoms n" in the target options.

Thanks,

- Lee-Ping