Abstract
Understanding the stability of the native state and the dynamics of a protein is of great importance for all areas of biomolecular design. The efficient estimation of the influence of individual contacts between amino acids in a protein structure is a first step in the reengineering of a particular protein for technological or pharmacological purposes. At the same time, the functional annotation of molecular evolution can be facilitated by such insight. Here, we use a recently suggested, information theoretical measure in biomolecular design โ the KullbackโLeiblerโdivergence โ to quantify and therefore rank residueโresidue contacts within proteins according to their overall contribution to the molecular mechanics. We implement this protocol on the basis of a reduced molecular model, which allows us to use a wellโknown lemma of linear algebra to speed up the computation. The increase in computational performance is around 10^1^โ to 10^4^โfold. We applied the method to two proteins to illustrate the protocol and its results. We found that our method can reliably identify key residues in the molecular mechanics and the protein fold in comparison to wellโknown properties in the serine protease inhibitor. We found significant correlations to experimental results, e.g., dissociation constants and ฮฆ values. ยฉ 2010 Wiley Periodicals, Inc. J Comput Chem, 2011