查看文章

Our current project [1] aims to develop high fidelity, computationally based, predictive mechanistic models of biomedical systems which can be applied in support of drug discovery and personalised medicine utilizing today’s top-level computational infrastructure. In the project, we investigate the robust application of free energy approaches in these two areas within a computing infrastructure of the highest performance. Therefore, this project will not only advance the particular fields in focus, but will lead to improved and novel insights for the operation of HPC machines at unprecedented levels, thereby serving as a lighthouse model for other domains. The theme of the project is gaining insight into the binding properties of proteins which represent key classes of drug target in important disease cases.

Over the past few years, we have uncovered and developed two new ways of calculating the free energy of binding of ligands to proteins. One is ESMACS (enhanced sampling of molecular dynamics with assumption of continuum solvent)[2]; the other is TIES (thermodynamic integration with enhanced sampling)[3]. We also investigate new approaches to enhance sampling and for more precise free energy estimations in an extension of the BAC workflow environment; the approaches include the most popular Hamiltonianreplica exchange (H-REMD) and its variants–replica exchange with solute tempering (REST2) and free energy perturbation with REST2 (FEP/REST2).