Title: Advances in free-energy calculations at the dawn of exascale computing
Speaker: Chris Chipot, Adjunct Professor, Department of Biological Physics, The Centre National de la Recherche Scientifique, Paris, France
In recent years, free-energy calculations have emerged as an indispensable tool to bridge the gap between molecular dynamics simulations and biological events, and help tackle deep biological questions that experiment alone has left unresolved. In spite of spectacular advances on the hardware front, which have paved the way for brute-force molecular dynamics to time and size scales hitherto never attained, free-energy calculations represent a cogent alternative to access with unparalleled accuracy the thermodynamics and possibly the kinetics that underlie complex processes of the living. In this lecture, I will review recent developments for free-energy calculations of both alchemical and geometrical transformations. In particular, I will show how combination of importance-sampling schemes can be used profitably to map with unprecedented efficiency and without any loss of accuracy complex free-energy landscapes from whence physically meaningful transition pathways can be extracted. I will compare this strategy with an alternative route consisting of a path search in multidimensional space followed by a potential of mean force calculation along the putative minimum free-energy pathway. I will also show how GPU-acceleration can be harnessed to boost free-energy calculations of an alchemical nature, and, thus, allow large sets of mutations to be performed in a routine fashion on affordable platforms equipped with GPUs. I will close discussing the hurdles free-energy calculations have overcome in recent years, and the challenges they are facing as we are at the dawn of exascale computing.