Parallel Flexible Molecular Docking in Computational Chemistry on High Performance Computing Clusters

摘要

The main objective in pharmaceutical research is development of novel drugs with improved biological effect in specifically afflicted organisms. A common practice in drug design focuses on systematic organic derivatization of chemical structures exhibiting certain biological activity and subsequent biological in vitro evaluation of the resulted benefits. However, this classical approach can be more or less classified as a chance drug discovery, being very arduous, expensive and time consuming. Nowadays, a lot of enthusiasm is given to rationally oriented drug research techniques like computer-aided drug design, virtual screening, bioinformatics, chemometrics, quantitative structure-activity relationships, etc. In the present article, we deal with designing a high performance computing (HPC) support for flexible molecular docking (FMD) which can be beneficially utilized in structure-based virtual screening (SBVS). The principles of FMD are briefly introduced and a solution combining message passing interface (MPI) with multithreading is proposed. The merits (e.g. availability, scalability, performance) of MPI-HPC enhanced SBVS/FMD are compared with other HPC techniques utilized for novel lead structures discovery in medicinal chemistry.