Ultra-high throughput vHTS with neural networks


Victor S. Lobanov, 3-Dimensional Pharmaceuticals, Inc, 665 Stockton Dr., Suite 104, Exton, PA 19341
3D pharmacophore- and docking-based virtual screening involves intensive computations that impose severe limitations on the number of compounds that can be screened within a reasonable period of time. In contrast, artificial neural networks trained with 1D and 2D molecular descriptors offer several orders of magnitude higher throughput and can be applied to screen truly massive collections. In this presentation, pros and cons of neural network based virtual screening techniques are discussed and examples of neural networks trained to predict activity and ADME properties are presented.