Time: 2:00-3:00pm

Venue: Lecture Hall, Jia Yi Bing Building, 82 Jing Chun Yuan, BICMR

Speaker: Anthony Bak, Stanford University and Ayasdi Co. Ltd.

Abstract: A basic problem in drug discovery is to search a large space of compounds to find those that have a high probability of curing disease. We show how to use topological descriptors, persistence diagrams formed with respect to both geometric and chemical properties of compounds, to organize the compounds into a metric space. In the particular case we'll discuss we localized known human cancer drugs within the space as well as separated out bio-activity by species. We reach state of the art classification accuracy while offering a "global view" of the compound space inaccessible before.