GRADIENT SHAKE - AN IMPROVED METHOD FOR CONSTRAINED ENERGY MINIMIZATION IN MACROMOLECULAR SIMULATIONS

Current macromolecular energy minimization algorithms become inefficie nt and prone to failure when bond length constraints are imposed. They are required to relieve steric stresses in biomolecules prior to a mo lecular dynamics simulation. Unfortunately, the latter often require c onstraints, leading to difficulties in initiating trajectories from un constrained energy minima. This difficulty was overcome by requiring t hat the components of the energy gradient vanish along the constrained bonds. The modified energy minimization algorithm converges to a lowe r energy in a fewer number of iterations and is more robust than curre nt implementations. The method has been successfully applied to the Di ckerson DNA dodecamer, CGCGAATTCGCG. (C) 1995 by John Wiley & Sons, In c.