Hi all,
I have conducted comparisons for different energy contribution in the
99SB/14SB and 19SB FFs using Amber and Gromacs. Between both engines, I
found most of the energies identical. However, I observed deviations in
the Coulomb SR and Coulomb 14 energies, and I’ve been wondering at the
reason.
*WHAT I DID * *
*
*
*
*In short: *
*I compared the energy contributions of 99SB/14SB and 19SB force fields,
calculated using Gromacs with topologies converted from Amber, with
those calculated using cpptraj. *
*Long **story**:*
For 99SB, 14SB, and 19SB, I built all topologies in Leap and employed
|amb2gro_top_gro.py|to convert them into Gromacs topologies. A 500 ns
simulation of a small protein (1770 atoms) was conducted in Gromacs, and
the trajectory was utilized for all subsequent calculations. Water and
ions were removed, and the trajectory was centered. In the Gromacs
calculation, I applied no cutoff for nonbonded interactions (all cutoffs
much larger than the system) and no neighbor list
(verlet-buffer-tolerance = -1, nstlist=1, rlist=50). The same xtc
trajectory was used for all calculations. All Amber calculations were
executed using cpptraj and its energy command. Various energies were
calculated for all three force fields in both Amber and Gromacs for
comparison.
*RESULTS*
*
*
The results, presented in two columns, display the time series on the
left and the distributions of energy differences on the right.
*QUESTION*
I)
I am attempting to interpret the differences in Coulomb energies. The
Coulomb short-range (SR) energies exhibit deviations of approximately 3
kJ/mol. Interestingly, when comparing the energies from this Gromacs run
with another Gromacs run where I built the topology internally (FF99SB),
the energies are identical. I verified the charges, and they are
consistent. Would you deem this energy difference tolerable? Is there a
possible explanation for this, such as variations in electrostatics
implementations?
II)
Concerning the disparities in Coulomb 14 interactions, I observed that
the difference decreases when directly comparing energies from cpptraj
with results from Gromacs using an internally built topology (FF99SB).
Could there be something missing in the topology converted from Leap
with amb2gro_top_gro.py? Is there any known information on this?
Thank you in advance!
Best
Albert
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Jan 26 2024 - 03:00:02 PST
