Alchemical Free Energy calculation (dH/d\lambda) of a tabulated potential

User discussions
1

GROMACS version: 2018.8
GROMACS modification: No

I am using tabulated potentials simulation to model lanthanides non-bonded interactions (12-6-4) form instead of standard (12-6) form.

I am now trying to calculate the solvation free energy of those lanthanides through fading them in or out of the solution using alchemical path.

However, I am not sure if gromacs handles non-standard tabulated potential energy function derivative with respect to lambda (dh/dl) correctly or not.

I’ve looked into the manual, and there was nothing about how dh/dl is calculated for tabulated non-bonded interactions (like vdw and coulomb).

My question would be, does gromacs know how to handle derivates with respect to lambda of tabulated non bonded potentials ?

If not, is there a way to compute solvation free energy in gromacs of those lanthanides described through those non-bonded tabulated potentials ?

Thank you,
Hossam

2

Hello Hossam,

I am facing the same question, did you find out anything related to your question? I want to calculate the SFE using a 16-6 Mie Potential?

3

This should be handled correctly.

4

Hello Hess,

I could not find documentation indicating that tabulated non-bonded interactions are supported under the free-energy code section of the manual. The manual provides detailed coverage for other interaction types, but I was unable to find similar guidance for tabulated non-bonded interactions.

Based on this, my understanding is that this interaction type is likely not supported within the free-energy code.

Best,
Hossam

5

Hello Luedsch,

Unfortunately, I ended up scripting my own thermodynamic path and performing the calculations manually using Python scripts I developed for this case.

I also decided not to compute absolute solvation free energies, because I did not want to implement a soft-core potential scheme myself. Instead, I focused on relative solvation free energies between two species, which was sufficient for my problem.

I had also hoped to use the lambda-ramping protocol for nonequilibrium work estimation, but I refrained for the same reason: I do not believe tabulated non-bonded interactions are properly supported in the free-energy code, particularly for soft-core treatment.

Best,
Hossam

6

Tabulated non-bonded interactions are supported in free-energy calculations (and GROMACS would exit with an error message if they would not).