The advent of largescale toxicogenomics databases based on <i>in vivo </i>studies allows direct association of histopathological phenotypes with gene expression across a range of time-points and doses. This association is not currently clear but is crucial in helping to understand potential mechanisms of toxicities in order to reduce late stage drug development failure due to unacceptable toxicity.<br/>In this work, histopathological signatures were generated from the DrugMatrix database, using the histopathology ontology, HPATH, to group similar assays. These distinct signatures provided the basis for generating consensus toxic networks that were investigated using weighted gene co-expression network analysis (WGCNA). As a case study, cellular infiltrate in the liver was selected, 25 instances with the same signature defined as the toxic set and a consensus co-expression network was generated. Analysis of modules, defined as highly correlated genes, revealed five modules that were not conserved between the toxic and the control cases. The OPEN TG-GATES database was used to validate the conservation of these modules and showed one to be highly conserved and annotated with an appropriate histopathology signature. This module was significantly enriched in genes involved in cytokine and drug response pathways. This was confirmed by annotating with the Comparative Toxicogenomics Database, revealing the consistency with current knowledge and allowing previously unstudied compounds to be classified. This work shows associations between highly correlating genes and toxicities. This provides the potential to identify early stage biomarkers for toxicity that would otherwise be found from late stage histopathology.<br/>

Determining associations between transcriptomics and toxicity using co-expression network methods

256th National Meeting (2018)

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