Comparative analysis of machine learning methods in ligand-based virtual screening of large compound libraries.

Machine learning methods have been explored as ligand-based virtual screening tools for facilitating drug lead discovery. These methods predict compounds of specific pharmacodynamic, pharmacokinetic or toxicological properties based on their structure-derived structural and physicochemical properties. Increasing attention has been directed at these methods because of their capability in predicting compounds of diverse structures and complex structure-activity relationships without requiring the knowledge of target 3D structure. This article reviews current progresses in using machine learning methods for virtual screening of pharmacodynamically active compounds from large compound libraries, and analyzes and compares the reported performances of machine learning tools with those of structure-based and other ligand-based (such as pharmacophore and clustering) virtual screening methods. The feasibility to improve the performance of machine learning methods in screening large libraries is discussed.

SVM model for virtual screening of Lck inhibitors.

Lymphocyte-specific protein tyrosine kinase (Lck) inhibitors have treatment potential for autoimmune diseases and transplant rejection. A support vector machine (SVM) model trained with 820 positive compounds (Lck inhibitors) and 70 negative compounds (Lck noninhibitors) combined with 65 142 generated putative negatives was developed for predicting compounds with a Lck inhibitory activity of IC(50) < or = 10 microM. The SVM model, with an estimated sensitivity of greater than 83% and specificity of greater than 99%, was used to screen 168 014 compounds in the MDDR and was found to have a yield of 45.8% and a false positive rate of 0.52%. The model was also able to identify novel Lck inhibitors and distinguish inhibitors from structurally similar noninhibitors at a false positive rate of 0.27%. To the best of our knowledge, the SVM model developed in this work is the first model with a broad applicability domain and low false positive rate, which makes it very suitable for the virtual screening of chemical libraries for Lck inhibitors.