Genetic refinement of the molecular force field: Normal mode analysis of purine vibrations

A new approach to the empirical force field determination by the adaptive optimization method is studied. The problem is conceived as a multiparameter optimization task and solved by the genetic algorithms (GA). The principles of the approach and the most efficient GA strategies are presented. With regard to the computational demands, the GA parameters were looked for in the test series in the force field refinement of the CHCl3 molecule. The most efficient ones were applied to the force field refinement of the bigger molecule of purine. The results show that the GA approach is powerful enough to provide a plausible force field of purine. The perspectives of the approach to bigger systems are outlined. © 1997 John Wiley & Sons, Inc. Biospect 3: 61–71, 1997