This talk discussed two applications of Monte Carlo methods to computational biology. The first application was the simulation of highly-constrained, cyclic peptides. Such molecules are used in nature as signaling molecules and in the pharmaceutical industry as initial trial structures for drugs. Professor Deem discussed biased Monte Carlo methods for simulating cyclic peptides. Backbone atoms are equilibrated with a biased rebridging scheme, and side-chain atoms are equilibrated with a look-ahead configurational bias Monte Carlo. Parallel tempering is an essential component of the method. The approach was illustrated on a variety of proline-containing, cyclic peptides that undergo cis/trans isomerization of the CONH amide bond. In the second application, Professor Deem discussed protein molecular evolution. A hierarchical approach to the efficient searching of protein sequence space was presented, derived by analogy with biased Monte Carlo methods. The evolutionary potential of the protocol on a model of protein sequence and function was then quantified. These simulations demonstrated that non-homologous juxtaposition of encoded structure is a rate-limiting step in the production of new tertiary protein folds. Non-homologous "swapping" of low-energy secondary structures increased the binding constant of a simulated protein significantly. Applications of the approach include the generation of new protein folds and modeling the molecular evolution of disease.
Professor Michael W. Deem
