Elucidation of differential response networks from toxicogenomics data

TL;DR

This paper introduces a new network-based method for analyzing toxicogenomics data, identifying drug-affected biological pathways and mechanisms that traditional gene expression analysis might overlook.

Contribution

The authors developed a novel approach that integrates network connectivity with gene expression data to elucidate differential biological pathways in response to drugs.

Findings

Identified drug-specific differential pathways with correlated expression patterns.

Reconstructed network modules that distinguish chemical treatments.

Demonstrated improved detection of cellular mechanisms over traditional methods.

Abstract

We describe a novel approach to the analysis of toxicogenomics data and elucidation of biological networks affected by drug treatments. In this method approximately 15,000 linear pathway modules were generated from manually assembled pathway maps from MetaCore (GeneGo, Inc.). Microarray expression data from livers of rat exposed to phenobarbital, mestranol and tamoxifen were mapped onto these modules. Using different analytical techniques we have identified sets of "differential" pathways featuring highly correlated expression among multiple repeats of the same treatment while showing strong anti-correlation across different treatments. Network modules distinguishing chemical treatments were re-assembled based on these pathways. Unlike traditional statistical and clustering procedures in expression profiling, our method takes into account both network connectivity and gene expression in the course of the analysis. We demonstrate that it enables identification of important cellular mechanisms involved in drug response that would have been missed by the analysis based on individual gene expression profiles.