T1000: A reduced toxicogenomics gene set for improved decision making
- Published
- Accepted
- Subject Areas
- Bioinformatics, Computational Biology, Toxicology, Data Mining and Machine Learning
- Keywords
- toxicogenomics, gene signature, co-expression network, graph clustering, machine learning, gene selection
- Copyright
- © 2019 Soufan et al.
- Licence
- This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. For attribution, the original author(s), title, publication source (PeerJ Preprints) and either DOI or URL of the article must be cited.
- Cite this article
- 2019. T1000: A reduced toxicogenomics gene set for improved decision making. PeerJ Preprints 7:e27839v1 https://doi.org/10.7287/peerj.preprints.27839v1
Abstract
There is growing interest within regulatory agencies and toxicological research communities to develop, test, and apply new approaches, such as toxicogenomics, to more efficiently evaluate chemical hazards. Given the complexity of analyzing thousands of genes simultaneously, there is a need to identify reduced gene sets.Though several gene sets have been defined for toxicological applications, few of these were purposefully derived using toxicogenomics data. Here, we developed and applied a systematic approach to identify 1000 genes (called Toxicogenomics-1000 or T1000) highly responsive to chemical exposures. First, a co-expression network of 11,210genes was built by leveraging microarray data from the Open TG-GATEs program. This network was then re-weighted based on prior knowledge of their biological (KEGG, MSigDB) and toxicological (CTD) relevance. Finally, weighted correlation network analysis was applied to identify 258 gene clusters. T1000 was defined by selecting genes from each cluster that were most associated with outcome measures. For model evaluation, we compared the performance of T1000 to that of other gene sets (L1000, S1500, Genes selected by Limma, and random set) using two external datasets. Additionally, a smaller (T384) and a larger version (T1500) of T1000 were used for dose-response modeling to test the effect of gene set size. Our findings demonstrated that the T1000 gene set is predictive of apical outcomes across a range of conditions (e.g.,in vitroand in vivo, dose-response, multiple species, tissues, and chemicals), and generally performs as well, or better than other gene sets available.
Author Comment
This is a submission to PeerJ for review.
Supplemental Information
Density plot of lactate dehydrogenase (LDH) activity (%) across human and rat in vitrohepatic experiments from the OPEN TG-GATEs Project
About 86% of experiments were indicated normal in the range of 95%-105% and the remaining 14% were cytotoxic cases. 95% and 105% are cut-offs that appear at 5% of left and right tails, respectively.
