High-Throughput Screening with Two-Dimensional Kernel Methods
Chloe-Agathe Azencott and Pierre Baldi
Institute for Genomics and Bioinformatics, Donald Bren School of Information and Computer Science
University of California, Irvine
cazencot@ics.uci.edu

High-throughput screening, or the rapid testing of thousands or millions of molecules to identify active compounds, is an important component of drug discovery. The ability to conduct these experiments in silico implies a more time-efficient and less costly process and is becoming indispensable in the pharmacology industry. In this talk we present two-dimensional kernels based on graph representations of molecular compounds and show how they can be used to derive a support vector machine classifier that is effectively retrieving active compounds in the HIVA library of the IJCNN Agnostic Learning vs. Prior Knowledge Challenge, with minimum training and limited over-fitting. Moreover, the kernels presented here are general enough to be applied to other problems in the chemistry domain, such as the prediction of molecular properties.