Title:  Integrating computational analysis and protein engineering to study myosin mechanics

Abstract:
Faced with the structural artifacts that are caused by the environment in which the structures are obtained, for example crystal packing conditions in X-ray crystallography. In order to examine whether standard structure based function prediction methods could benefit from examining conformational space of the molecules and evaluating many structures for each molecule, we coupled molecular dynamics simulations, which generate structural diversity, with FEATURE[1], a machine learning Bayesian algorithm that identifies sites of interest based on evaluating physico-chemical 3D environments, and examined FEATURE’s effectiveness in finding Ca2+ binding sites.