|
Abstract:
|
The modern drug discovery process involves multiple sources of high-dimensional data. A typical example is the integration of chemical structure, bioassay read-outs, and transcriptomic (gene expression) data to better understand the biological mechanisms of candidate drugs enabling early detection of safety issues prior to later and expensive phases of drug development cycles. Our goal is to link between the transcriptomic and the phenotypic data of compounds taking into account that a chemical structure may influence both biological variables. In this paper, a structure specific and gene-specific joint model is presented as a tool to model the association between gene expression and bioactivity (measured by pIC50) adjusting for a chemical structure. Two types of gene-bioactivity association are of interest: (1) an association which is induced by a chemical structure and (2) an association that reflects the correlation between gene and pIC50 facilitating the identification of genes that can serve as biomarker for compounds efficacy. Two oncology projects are used to illustrate the applicability and usefulness of the joint modeling approach to aid drug discovery.
|
