OpenMolGRID, a GRID Based System for Solving Large-Scale Drug Design Problems

In our days, pharmaceutical companies are screening millions of structures in silico. These processes require fast and accurate predictive QSAR models. Unfortunately, at the moment these models do not include information-rich quantum-chemical descriptors, because of their time-consuming calculation procedure. These challenges make indispensable the usage of large-scale QSAR calculations on GRID systems. These “high-throughput” informatics systems provide the facility to develop QSAR models on a vast number of model compounds in a short time and also the fast application of the novel method on an unprecedently high number of molecules. OpenMolGRID (Open Computing GRID for Molecular Science and Engineering) is going to be one of the first realizations of the GRID technology in drug design. Based on security considerations and its easy plugin technology UNICORE was selected as Grid middleware for the system (for additional info see the contribution of Mathilde Romberg et al., titled ‘Support for Classes of Applications on the Grid’). The OpenMolGRID system is designed to build forwardand reverse-QSAR models based on thousands of different type of descriptors, many of them requiring computation intensive 3D calculations. For reallife testing purposes, 30,000 novel and diverse structures have been synthesized and IC50 values for in vitro human fibroblast cytotoxicity (i.e. the concentration of compound that kills the 50% of the cells) is being determined. Using these experimental data, linear and non-linear forward-QSAR models is being developed and the predictive capability of these models is going to be validated. The system is capable to build and apply reverse-QSAR models, to find novel structures with favourable properties. The structural ideas are generated by a molecular engineering tool using specified chemical connectivity rules, and the reliable candidates are selected based on additive fragment descriptors defined for each building block. Using its implemented data warehouse technology, the system is suitable to collect data from geographically distributed, heterogeneous resources (for additional details see the contribution of Damian McCourt et al., titled ‘The OpenMolGRID Data Warehouse, MOLDW’). The system is going to analyze millions of structures in a considerable time using the obtained QSAR/QSPR equations with traditional and GRID based computation procedures, and select the most promising hits from this huge virtual chemical domain. OpenMolGRID enables highly specialized experts from several continents to make immediate decisions in the library design phase during a net meeting. The development of OpenMolGRID is partly funded by the European Commission under the 5th Framework Project (IST-2001-37238).