It was possible to build an amber topology file and convert it to gromacs. Here's what I knew in 2015 (I don't use gromacs anymore). You'll need to validate your converted topology, as I've seen months of simulations wasted. For sure don't use amb2gmx.pl
or ACPYPE. Marko Wehle's script (see attached) did work in 2012, but I do not recommend anyone does this. Gromacs is constantly changing and Marko's script is no longer being maintained. Further, the whole process adds an extra layer of complexity, increasing your chances of getting garbage. I'm attaching Marko's scripts here anyway, but you are on your own! Gromacs is faster when running on many cpus (e.g 128), but nowadays (2015) you'll get similar/better speeds when running Amber on modern gpus. Google "amber gpu benchmarks". Even if you are stuck with cpus, I still recommend amber as the speedup isn't worth the high risk of there being a mistake when you convert.
Script details from Marko: Note you need GROMACS > 4.5.x
works for both GLYCAM and AMBER topologies, including mixed 1-4 scaling, but writes dihedrals out in type 9 functions format and impropers for type 4 functions in GROMACS>4.5.x.
Attached you find the latest version, with corrections for box parameters and posre-itp hints, TIP3P water is now "hard" included, not in extra files anymore.
So it is easier to use already solvated systems from AMBER in GROMACS.
# Please cite the original source:
# Sorin & Pande (2005), Biophysical Journal, 88, 2472-2493.
# DePaul, Thompson, Patel, Haldeman, & Sorin (2010), Nucleic Acids Research, 38, 4856-4867.
# And please cite Marko's work, using the modifications:
# J. Am. Chem. Soc., 2012, 134 (46), pp 18964-18972
# Mechanical Compressibility of the Glycosylphosphatidylinositol (GPI) Anchor Backbone Governed by Independent Glycosidic Linkages
# Marko Wehle, Ivan Vilotijevic, Reinhard Lipowsky, Peter H. Seeberger, Daniel Varon Silva, and Mark Santer