Modeling Sidechains in Protein Folding
Matt J. Carlson, William A. Goddard III
In a hierarchical method of protein fold prediction, the final stages involve simulation and analysis of an all-atom representation of a protein structure including its sidechain atoms. However, current protein structure prediction methods in our group rely on a reduced representation of the protein comprised of only C? atoms or only main chain atoms. It is believed that a significant portion of the folding stabilization is a result of favorable packing of hydrophobic sidechains. Therefore, it is important to develop methods that can quickly and correctly place sidechain atoms onto a folded main chain backbone. Once an all atom protein model is constructed, further simulations and analysis can take place to help discriminate between prediction candidates. Current research has improved the placement of sidechain atoms through monte carlo techniques by increasing the number of sidechain conformations searched, but using a simple energy function to discard poor sidechain interactions. Our method focuses on fewer sidechain conformations, but a more complex representation of the sidechains. This allows near-native conformations that are slightly disfavored that would be discarded in other methods to be preserved. Our method is comparable to other current methods, but improvement is expected with the inclusion of solvent effects into the energy function.