Accelerating atomistic simulations of proteins by Bayesian inference with unreliable information;
Cell biology is sometimes cell physics
MIT Building 4, Room 237
Molecular simulations give insights and quantitation to protein folding, drug discovery and the binding of ligands, and biological mechanistic actions in the cell. But, even with current sampling methods, such as Replica Exchange, physical simulations are much too slow, and don't scale well to larger proteins or larger actions. We have developed MELD, which melds together vague external knowledge to accelerate physics-based molecular simulations. I will describe proofs of principle in folding proteins in the blind prediction event called CASP, and in computing binding affinities of peptide ligands to proteins.
Some behaviors of cells are not due to single proteins or pathways, but are due to the physical properties of proteomes as a whole. For example, the growth rates of bacteria as a function of temperature or salt can be explained the folding stability and diffusion rates of the proteins in the proteome. Using simple physical models, we explore physical aspects of cell mechanisms and evolution, also including cellular energy balance and proteostasis, the machinery that folds and disaggregates proteins.
Nanoscale Disorder Drives the Dynamics of Excitons in Molecular Semiconductors; What Can Interfacial Water Molecules Tell Us About Solute Structure?
MIT Building 4, Room 237
Many organic electronic materials are composed of soft condensed matter that is both electronically active and disordered on the nanoscale. The electronic properties of these materials can depend sensitively on the details of molecular morphology, reflecting a complex coupling between excited electrons and the disordered nuclear environment. To better understand this coupling and how nanoscale disorder affects the electronic dynamics in these materials we utilize numerical simulation. In this talk I describe our approach to unraveling the effects of nanoscale disorder on the dynamics of excitons, which utilizes atomistic simulation, coarse-grained models, and quantum dynamics.
The molecular structure of bulk liquid water reflects a molecular tendency to engage in tetrahedrally coordinated hydrogen bonding. At a solute interface water’s preferred three-dimensional hydrogen bonding network must conform to a locally anisotropy interfacial environment. Interfacial water molecules adopt configurations that balance water-solute and water-water interactions. The arrangements of interfacial water molecules, therefore encode information about the effective solute-water interactions. This solute-specific information is difficult to extract, however, because interfacial structure also reflects water’s collective response to an anisotropic hydrogen bonding environment. Here I present a methodology for characterizing the molecular-level structure of liquid water interface from simulation data. This method can be used to explore water’s static and/or dynamic response to a wide range of chemically and topologically heterogeneous solutes such as proteins.