One of the long-standing problems in shape searching has been the assignment of significance or probability to a given shape similarity between a query and a database molecule. This leads to uncertainty in which/how many molecules to select by shape similarity from a shape-based virtual screening experiment with the highest probability of success/lowest false positive rate. In this paper we will present the positive False Discovery Rate (pFDR) as a mechanism to address this problem. We will show how to arrive at an estimate of pFDR for queries of given properties and illustrate the application of pFDR to a variety of virtual screening scenarios. The pFDR approach, while developed here for ligand-centric methods, is completely general to determination of significance in many areas of computational chemistry.

False discovery rate: A “New Deal” for shape searching?

244th National Meeting (2012)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.