A collaboration of Asian and European laboratories has used the Enabling Grids for E-sciencE (EGEE) Grid infrastructure to analyse 300,000 possible drug components against the avian flu virus H5N1. The goal was to find potential compounds that can inhibit the activities of an enzyme on the surface of the influenza virus, the so-called neuraminidase, subtype N1. Using the Grid to identify the most promising leads for biological tests could speed the development of the drugs.
The challenge of this in silico drug-discovery application is to identify molecules that can dock on the active sites of the virus to inhibit its action. During four weeks in April, scientists used 2000 computers – equivalent to 100 years on a single computer – to investigate the docking of 300,000 compounds against eight structures of influenza A neuraminidases. More than 60,000 files with a data volume of 600 GB were created and stored in a relational database. Researchers are now identifying and ranking potential compounds according to the binding energies of the docked models. With the results from the in silico screening, they can predict which compounds and chemical fragments are most effective for blocking the active neuraminidases in case of mutations.
Using experience from a data challenge on malaria (WISDOM), the Grid-enabled in silico process was implemented in less than a month on three different Grid infrastructures, AuverGrid, EGEE and TWGrid, paving the way for a virtual large-scale drug-screening service. The WISDOM platform performed most of the computing. However, a lightweight application framework called DIANE (see CERN Courier September 2006 p20) was also adopted in this challenge and used to perform much of the activity to enable efficient computing-resource integration and usage.
Author:
Compiled by Hannelore Hämmerle and Nicole Crémel