Development of genetic biomarkers in drug discovery and early drug development experiments

Abstract

Quantitative Structure Transcriptional Activity Relationship (QSTAR) involves relating the three data sources in early drug discovery namely (1) chemical structure (fingerprint features), (2) bioactivity (bio-assay read-outs) data for targets of interest, and (3) transcriptomic (gene expression) data of a set of compounds. In this paper, a gene-specific and fingerprint feature-specific joint model is presented as a tool to model the association between gene expression and biological activity taking into account the chemical structure of the compounds. The model allows to detect genes that are associated to the bio-assay read-out for which some of the associations are mainly induced by certain fingerprint feature(s) of compounds. The joint model is applied to two oncology projects. Results show that a number of compounds’ fingerprint features have differential effects on both bio-assay read-outs and a set of correlated genes.