Information theoretical measures to analyze trajectories in rational molecular design.

We develop a new methodology to analyze molecular dynamics trajectories and other time series data from simulation runs. This methodology is based on an information measure of the difference between distributions of various data extract from such simulations. The method is fast as it only involves the numerical integration/summation of the distributions in one dimension while avoiding sampling issues at the same time. The method is most suitable for applications in which different scenarios are to be compared, e.g. to guide rational molecular design. We show the power of the proposed method in an application of rational drug design by reduced model computations on the BH3 motif in the apoptosis inducing BCL(2) protein family. (c) 2007 Wiley Periodicals, Inc.