[gmx-users] fABMACS: a new tool for free energy calculations

[Parker de Waal]

Hello GROMACS Users and Developers,

We'd like to bring your attention to a newly developed tool for free energy calculations, fABMACS. fABMACS is an efficient implementation of a mollified adaptive biasing potential that can be used to compute the free energy landscape for problems such as ligand binding or protein conformational changes.

To demonstrate significant improvements in computational efficiency, we benchmarked fABMACS against WT-metaD (using PLUMED2 as well as native implementations within AMBER16 and NAMD2.11) for systems of 35,000 and 99,000 atoms across a range of compute nodes. Compared to native GROMACS, fABMACS incurred ~5-9% overhead while PLUMED2 saw ~45-50% overhead for these systems making fABMCAS the fastest adaptive bias potential implementation to date.

fABMACS currently supports RMSD and dihedral collective variables (CVs), however we have plans to include distance CVs as well in the near future.

The fABMACS manuscript can be found here, http://scitation.aip.org/content/aip/journal/jcp/145/15/10.1063/1.4964776 , and our github page with installation guides and usage tutorials can be found here https://github.com/BradleyDickson/fABMACS

If you have any questions, suggestions or just want assistance setting up your own systems, please don't hesitate to ask us via email.

Happy simulating!
The fABMACS Team