We present a tutorial for the TDT challenge 1 in the form of an IPython notebook. IPython is a web-based interactive environment for creating �executable documents� that combine code, text, and graphics. It is accompanied by Python scripts for the computationally expensive parts. We start by filtering the HTS hit list for undesirable physicochemical properties and problematic substructures. A subset of the remaining compounds is selected for testing in a confirmatory assay by picking a diverse set of small compound clusters. Next a set of machine-learning methods and molecular fingerprints are evaluated using the HTS data. The three best models are combined using heterogeneous classifier fusion to further enhance the predictive power. This fusion model is used to rank-order the external test set. Finally, a library of commercially available compounds is ranked using the fusion model and a diverse set of small clusters is selected for confirmatory screening.

Antimalaria hit finding using classifier-fusion boosted predictive models

248th National Meeting (2014)

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