Configuration and Deployment of a Virtual Cluster for Molecular Docking Experiments on the PRAGMA Cloud
Karen Rodríguez, Kevin Lam, Jason Haga, Kohei Ichikawa · 2013
Molecular docking simulations are an important tool that is often employed in biomedical research and pharmaceutical industries to scan compound databases in pursuit of a specific molecular interaction. This is often used in drug screening experiments when an entire drug compound database cannot realistically be tested with assays on a wet laboratory bench. DOCK is a computer application developed to aid researchers by simulating molecular bindings between a single cellular receptor and a list of ligands. DOCK then calculates an estimate of the stability of the bond resulting from a number of conformations for each ligand. This allows one to obtain a reduced list of compounds that are more likely to interact with the receptor of interest in a desired way in real life. The result would narrow down experiments to only include compounds expected to bind to a receptor in a certain way; thus allowing scientists to finish such experiments within a reasonable time and economic cost. DOCK, however, is a demanding program in terms of processing power. Additionally, this is known to not yield consistent results when executed from machines with varying processors, operating systems, and compilers. Availability of a reliable environment on which DOCK can be executed and yield reliable and quick results is therefore very important for large-scale experiments of drug screening nature. The purpose of this project was to create such environment out of networked virtual machine (VM) clones and upload it onto the PRAGMA cloud in order to meet DOCK’s high computational power demands whilst keeping computation times