Difference in Generalized Born energy

[tejtaruns]

Hi,
I am using the GBNeck2 model for implicit solvent in openMM and in Amber. But, I am getting a different energy of the generalized born term in openMM and in Amber.

Is it due to different non-polar models used by Amber and openMM?

[peastman]

Possibly? They use different approximations for the solvent accessible surface area. But without knowing exactly what you're doing, it's impossible to know what else might be different in your simulations. Can you provide more details, preferably a complete test case that shows what you're doing?

[tejtaruns]

Just for testing purposes, I am calculating single-point energy for a protein using Amber's ff14SB forcefield in the implicit solvent model(GBNeck2) in openMM as well as in Amber.
For openMM, I found forcefield files in openmmforcefields. And I am printing different energy contributions using forcegroups.
In Amber, I am using leaprc.protein.ff14SB and mbondi3 radii set in tleap. And for calculating single-point energy, I am using the igb=8 flag in the minimization input file.

And after this exercise, I am finding that every other contribution to energy is the same except the generalized born term.

[peastman]

The GB energy is proportional to the charge, while the SA energy is independent of it. Try setting all charges to 0 and see how the energy compares to Amber. Any difference is coming solely from the SA term. And if it's the same as before, you know there's no difference in the GB part.