Reparametrization of Protein Force Field Nonbonded Interactions Guided by Osmotic Coefficient Measurements from Molecular Dynamics Simulations

• DOI: 10.1021/acs.jctc.6b01059
• Publication Date: 2017-03-15T20:54:31Z
• Authors: Mark S. Miller; Wesley K. Lay; Shuxiang Li; William C. Hacker; Jiadi An; Jianlan Ren; Adrian H. Elcock
• Keywords: Force Field; Osmotic coefficient; Osmotic pressure; Parametrization (atmospheric modeling)

There is a small, but growing, body of literature describing the use osmotic coefficient measurements to validate and reparametrize simulation force fields. Here we have investigated ability five very commonly used field water model combinations reproduce coefficients seven neutral amino acids small molecules. The fields tested include AMBER ff99SB-ILDN, CHARMM36, GROMOS54a7, OPLS-AA, with first these in conjunction TIP3P TIP4P-Ew models. In general, for both molecules, produce computed that are lower than experiment; this indicative excessively favorable solute–solute interactions. sole exception general trend provided by GROMOS54a7 when applied acids: case, consistently too high. Importantly, show all can be made accurately experimental minor modifications–some previously reported others some new study–are van der Waals interactions charged terminal groups. Special care required, however, simulating Proline number fields, hydroxyl-group specific modification required order correct Serine Threonine simulated ff99SB-ILDN. Interestingly, an alternative parametrization latter field, proposed Nerenberg Head-Gordon groups, shown immediately excellent agreement experiment. Overall, study reinforces idea identify shortcomings fields' descriptions further demonstrates modifications parameters provide simple route optimizing