| Molecular recognition, especially in proteins and nucleic acids, is at the heart of much of modern chemistry and biology. We present here the diverse uses of Monte Carlo methods to study how biological molecules recognize each other at different levels of details. Monte Carlo methods are a class of computational algorithms that rely on repeated random sampling to compute their results. We first describe the implementation of a general and flexible Monte Carlo module for the program CHARMM, which is used widely for atomic-resolution modeling of biomolecular systems. Its application to the sampling of the configuration spaces of two peptides demonstrates the attractiveness of Monte Carlo for biomolecular simulations. Then, we present bioinformatic studies. A Gibbs sampling approach, based on the principle of Monte Carlo, is employed to predict transcription factor binding sites from sequence data obtained from high-throughput protein-DNA interaction mapping techniques. The results show that the method can predict binding motifs with no prior knowledge and is efficient compared with existing methods. The revolutionary generalization of Monte Carlo to the sampling of dynamical trajectories in complex systems, transition path sampling, is examined subsequently. We introduce a method, in the context of transition path sampling, for generating reactive trajectories when an existing one is not already available. Application to basin-to-basin hopping in a two-dimensional model system demonstrates that the method can yield unbiased reactive trajectories in an order of magnitude less computer time than competing methods. Application of the method and other recent advances in computational approaches for studying dynamics in complex 1 systems enables the first direct study of how a protein that maintains the human genome detects DNA damage. The success of Monte Carlo methods in different contexts suggests the importance and promising capabilities of Monte Carlo in future computational studies of biological systems. |
