Virtual Screening for Drug Discovery using Cloud

Damjan Temelkovski, King’s College London / Francis Crick Institute

Molecular docking is a computer simulation that predicts the binding affinity between two molecules, a ligand and a receptor. Large-scale docking simulations, using one receptor and many ligands, are called structure-based virtual screening. Often used in drug discovery, virtual screening can be very computationally demanding. This is why web or desktop applications that enable running simulations on powerful computing infrastructures (e.g. HPC) have been created. We have extended a desktop application for virtual screening to use cloud computing with an approach based on scientific workflows. In a proof of concept the Raccoon2 desktop application, WS-PGRADE workflows, and gUSE services with the CloudBroker Platform were used. This environment has enabled us to run large-scale simulations using a receptor of a protozoan parasite that causes the STI trichomoniasis, and 130216 candidate drugs. We show that the same approach can be used on private and public clouds and provide evidence for the optimal cloud configurations for virtual screening. This approach enables scientists to use their favourite desktop environment and provides them resources without major capital investment