Cloud computing for fast prediction of chemical activity

Quantitative Structure-Activity Relationships (QSAR) is a method to create models that can predict certain properties of compounds. Because of the importance of QSAR in designing new drugs, ability to accelerate this process becomes crucial. One way to achieve that is to be able to quickly explore the QSAR model space in the search for the best models. The cloud computing paradigm very well fits such a scenario, thus we designed and implemented a tool for exploration of the model space using our e-Science Central platform supported by the cloud. We report on scalability achieved and experiences gained when designing this system. The acceleration obtained is much beyond what existing QSAR solutions can offer, which opens potential for new interesting research in this area. Quantitative Structure-Activity Relationships, machine learning, cloud computing, scalability, performance evaluation