Don't miss the forest for the trees: multiscale modeling of biological systems

AVILA CL

San Javier

Congreso; XLI Reunion Anual de la Sociedad Argentina de Biofisica; 2012

Sociedad Argentina de Biofisica

Biological processes consist of an interplay of phenomena taking place at different time and length scales. Processes such as electron transfer or ligand binding to a protein might result in a conformational change triggering a signaling pathway or protein aggregation. Our understanding of biological processes has been aided by the use of computational models. Several computational methodologies have been developed for modeling the systems introducing increasing level of simplification, going from quantum mechanics, through all-atom, coarse-grained up to mesoscopic and mean field descriptions. Although the processes at each scale are intimately connected to one another, trying to embrace the system as a whole with a unique, minute detailed technique is neither possible nor desirable. In order to tackle this problem in an integrative way we must resource to a hierarchical approach. Multiscale approaches have arisen as a solution to bridge the gap between different levels of theory. The idea behind multiscaling is to use coarse models to enable quick and enhanced sampling of the phase space and move back to finer, more accurate resolutions in the regions of interest. For this we need physically relevant coarse models to correctly sample meaningful regions of the phase space and a proper methodology to transfer information between the finer and coarser models. While algorithms for information exchange between models are a field of active research, some of them have been successfully applied for bridging quantum mechanics and all-atom models to describe chemical reactions within enzymes. More recently with the advent of new coarse-grained models for proteins, DNA and lipids a lot of effort has been put into bridging coarse grained with all-atom models. I will briefly review on different approaches available to combine coarse-grained and all-atom simulations including the one used in our group, i.e. the use of reference potentials. I will also report on our advances in the development of a coarse-grain model based on the Amber all-atom force-field. I will illustrate the use of the methodology in the prediction of hotspots residues at protein-protein interfaces.