posted Aug 24, 2016, 12:24 PM by Emilio Gallicchio
updated Aug 24, 2016, 12:31 PM
The posters by Baofeng Zhang and Rajat Pal were selected for the prestigious interdisciplinary Sci-Mix event at the Fall ACS meeting in Philadelphia :
Targeting viral receptors using binding free energy-based virtual screening and GPU-accelerated software
Baofeng Zhang et al.
Abstract:We will describe the GPU-accelerated formulation of our single-decoupling free energy perturbation method for the rapid estimation of absolute binding free energies of protein-ligand complexes. The computational system is based on the OpenMM MD engine and the Asynchronous Replica Exchange job coordination system. While computationally intensive, free energy-based virtual screening can more accurately discriminate binders from non-binders because it includes entropic and conformational reorganization effects in addition to energetic complementarity factors. The method is therefore suitable for the secondary screening of focused ligand libraries, for example to improve on docking enrichment. Because it targets absolute binding free energies, this protocol is useful for preliminary free energy scoring of ligand libraries with diverse scaffolds. We will present examples drawn from recent screening campaigns targeting viral targets with alpha-hydroxytropolone derivatives. The combination of binding free energy predictions with structural analysis of dynamical trajectories has pointed out key aspects which we are currently pursuing to optimize leads and identify novel classes of inhibitors.
Alchemical computational methodologies for the estimation of binding free energies of supramolecular complexes
Rajat Kumar Pal et al.
Abstract:We will review the development and current applications of computational binding free energy protocols based on the alchemical Single-Decoupling Method (SDM) and multi-dimensional Asynchronous Replica Exchange sampling framework (AsyncRE). Asynchronous equilibrium techniques of this kind are particularly suited for heterogeneous computational clusters and computational grids. We have started to explore non-equilibrium approaches which, although inherently less statistically efficient than equilibrium ones, have the advantage of not incurring the overhead of communication between computational threads in a dynamic processing environment. We will describe a novel asynchronous methodology for binding free energy estimation which works under non-equilibrium conditions and requires no exchanges between replicas, thereby leading to higher computational throughput than equilibrium approaches on computational grids. We will present illustrative examples of equilibrium and non-equilibrium calculations drawn from recent investigations of Octa-acid and CBClip host-guest supramolecular complexes. The results indicate that non-equilibrium techniques have the potential to accelerate and streamline binding free energy calculations on computational grids.