Nowadays research community has moved from the COMPASS force fields to OPLS for the MD simulations of polymers. Is that just due to change in the Dihedral and harmonic terms?
In short, what are the advantages of OPLS-AA over COMPASS?
OPLS-AA has much simpler force field form than COMPASS, and the parameters are completely public, therefore OPLS-AA can be employed in broad range of MD programs; in particular, the OPLS-AA has already been embedded in the most popular MD program GROMACS, so utilizing OPLS-AA forcefield for polymer simulation is rather straightforward. In contrast, the COMPASS forcefield has much complicated form and thus difficult to be implemented in most general MD programs; in addition, parameters of COMPASS are private in the commerical Material Studio package and not transparent to users. These reasons severely limited the use of COMPASS for molecular modeling community. However, if you already have MS licence, using COMPASS is a good choice.
The OPLS philosophy is based on fitting MM PES surfaces (parameters) to experimental data sets, which render this FF especially powerful for accurately describing condensed phase properties of small drugs.