Title: Computational Tools to Study Protein Low-Frequency Motions
Folded proteins are not rigid objects. Their motions occur at many different frequencies, but the low-frequency motions are often the most pertinent for the protein functions (binding). In this talk, I will review several computational tools that we are developing to efficiently model, analyze, and predict protein low-frequency motions. I will present past and ongoing work along three different lines of research. (1) I will present new techniques to determine structural heterogeneity in folded proteins from X-ray crystallography data. These techniques produce a small collection of conformers (discrete sub-states) that best explain a given electron-density map, rather than a unique "average" conformer with large harmonic thermal noise. (2) I will describe geometric/kinematic tools to efficiently sample conformations of folded proteins or protein fragments; I will illustrate their application to the recognition of calcium-binding loop conformations. (3) Finally, I will discuss the generation of Hidden Markov models to compactly represent the long timescale dynamics of proteins among meta-stable states. In particular, I will discuss tools to score such models.