Support vector machines in combinatorial chemistry

The application of support vector machines (SVM) in a combinatorial drug design process was discussed. The SVM is a supervised machine learning technique that minimizes a bound on the expected generalization error by minimizing the composite error. A structure-activity relationship (SAR) analysis was performed in the drug discovery process to classify the suitability of the new molecular combinations. The SVM outperformed four frequently used techniques in a trial on data provided by GlaxoSmithKline Pharmaceuticals where it showed a high accuracy in classifying the more important of the two compound classes.