American Chemical Society
jp2068123_si_001.pdf (557.36 kB)

Efficient Calculation of Molecular Configurational Entropies Using an Information Theoretic Approximation

Download (557.36 kB)
journal contribution
posted on 2012-03-08, 00:00 authored by Bracken M. King, Nathaniel W. Silver, Bruce Tidor
Accurate computation of free energy changes upon molecular binding remains a challenging problem, and changes in configurational entropy are especially difficult due to the potentially large numbers of local minima, anharmonicity, and high-order coupling among degrees of freedom. Here we propose a new method to compute molecular entropies based on the maximum information spanning tree (MIST) approximation that we have previously developed. Estimates of high-order couplings using only low-order terms provide excellent convergence properties, and the theory is also guaranteed to bound the entropy. The theory is presented together with applications to the calculation of the entropies of a variety of small molecules and the binding entropy change for a series of HIV protease inhibitors. The MIST framework developed here is demonstrated to compare favorably with results computed using the related mutual information expansion (MIE) approach, and an analysis of similarities between the methods is presented.