Performance of four different force fields for simulations of dipeptide conformations: GlyGly, GlyGly-, GlyGly · Cl-, GlyGly · Na+ and GlyGly · (H2O)2.

Monte-Carlo conformational searches with four kinds of force fields (AMBER94, MM3*, MMFFs, and OPLS-2005) were performed on glycylglycine (GlyGly), deprotonated glycylglycine (GlyGly(-)), glycylglycine chloride anion complex (GlyGly · Cl(-)), glycylglycine sodium cation complex (GlyGly · Na(+)) and glycylglycine dihydrate [GlyGly · (H2O)2]. Combined with Hartree-Fock (HF) and second-order Møller-Plesset (MP2) optimizations, conformations within an energy of 20 kJ mol(-1) were predicted. After MP2 calculations, the geometries and relative energies of the predicted structures were the same regardless of the force field used. Therefore, the performance of different force fields reflects mainly the conformational search process. For GlyGly, there was practically no difference among the four force fields. Due to the complex hydrogen bonding network when involving water, the total number of resulting conformers for GlyGly · (H2O)2 increased drastically. Moreover, the MMFFs force field fared best in finding the global minimum compared to the remaining three force fields. In describing hydrogen bonded and inter-molecular complexes, we recommend application of the MMFFs and AMBER94 force fields. Furthermore, the MMFFs and OPLS-2005 force fields have a good description of electrostatic interactions. This work will contribute to helping the reader make an optimal choice of force field, taking into account the latter's strengths and limitations.