Protein conformational spaces: sampling, measuring and clustering

Proteins are large molecules composed of linear chains of connected aminoacids, being largely responsible for the processes taking place in living organisms. The conformation (spatial structural arrangment) adopted by those chains ensures the chemical and physical properties required for proper protein functioning, with a loss of the normal conformational features leading to malfunction or disease. Therefore, a detailed characterization of the distribution of conformations is determinant to understand and rationalize the behavior of proteins. This talk discusses some aspects and open issues in the study of protein conformational behavior in the field of Molecular Modelling. Although deterministic or stochastic computational methods can be used to sample protein conformations, the sampling is often partial and many alternative dissimilarity measures and classification algorithms exist. Thus, questions remain about the suitability of available measures, conformational spaces and clustering methods for properly reflecting the underlying energetics of these molecules and, ultimately, their thermodynamic and kinetic properties. Overall, the viewpoint adopted in the talk is that of a physical biochemist trying to take advantage of mathematical methods (and hoping to get helpful suggestions from the audience).