Biotechnology HPC Software Applications Institute (BHSAI)

BHSAI Scientists Mine Big Data to Identify Sets of Genes Predictive of Liver and Kidney Injuries

June 28, 2019   |  Download PDF

The goal of predictive toxicology and ultimately personalized medicine is to use your own cells to customize the best treatment for you. For this to work, test tube results (in vitro) need to have some connection with the toxic response in the human body (in vivo) so that the physician can correctly interpret the results and start treatment.

Developing a prediction tool is a great challenge, requiring artificial intelligence (AI) from both in vitro and in vivo experiments. Ideally, these experiments would be conducted in humans. But ethically, this is impossible. So we rely on the next best thing: animal experiments.

A computational approach to bridge in vitro and in vivo experiments, both within and across species. Data from Toxicogenomics Project-Genomics Assisted Toxicity Evaluation System (TG-GATEs) were used to identify key genes (modules) associated with liver injuries in rats in vivo, which were validated and further tested in rat in vitro and in human in vitro studies.A computational approach to bridge in vitro and in vivo experiments, both within and across species. Data from Toxicogenomics Project-Genomics Assisted Toxicity Evaluation System (TG-GATEs) were used to identify key genes (modules) associated with liver injuries in rats in vivo, which were validated and further tested in rat in vitro and in human in vitro studies.


The first records of animals being used as models of human anatomy can be traced to Alcmaeon of Croton and Aristotle, in ancient Greece. Since then, animals have been used for centuries to predict how chemicals and environmental factors might affect humans. But comparing species is complicated because their genetics and bioavailability differ. Not surprisingly, many compounds may be toxic for one species, yet benign for others. For example, theobromine, the compound that gives chocolate its bitter taste, is enjoyed by humans but can be lethal for cats and dogs.

Whether or not animal models are useful in predicting how people respond to a chemical has recently been debated in the scientific community. Regardless of the answer, we are left with the problem of prediction. As U.S. Secretary of Health and Human Services Mike Leavitt stated in 2007, “Currently, nine out of ten experimental drugs fail in clinical studies because we cannot accurately predict how they will behave in people based on laboratory and animal studies.”

Given these shortcomings of animal studies, and the growing consensus that conducting such studies on a large scale or for routine experiments is unethical, much research has focused on developing cell-based (in vitro) assays as substitutes. But biological markers of cell toxicity are usually quite crude and only indicate a generic response, such as loss of cellular integrity or cell death.

The predictive toxicology program, led by Dr. Anders Wallqvist, is funded by the Defense Threat Reduction Agency (DTRA) and involves a collaboration between scientists from TATRC’s Biotechnology High Performance Computing Software Applications Institute (BHSAI) and Vanderbilt University. Under this program, Dr. Schyman, a scientist at BHSAI, has been working on a computational approach to elucidate how to connect the dots between in vitro and in vivo experiments and between results from different species.

Their idea was to identify groups of genes (gene modules) commonly associated with a specific type of organ injury. The scientists first identified injury-specific gene modules by applying hierarchical clustering and artificial intelligence techniques on large sets of rats in vivo data obtained from a public database [Open Toxicogenomics Project-Genomics Assisted Toxicity Evaluation System (TG-GATEs)] (Te et al., 2016). The BHSAI team identified 8 and 11 gene modules specific to liver and kidney injuries, respectively.

In a recent publication, the team successfully validated the injury-specific gene module approach by exposing Sprague-Dawley rats to thioacetamide, a known liver toxicant, by showing that the genes associated with liver injury could predict the organ injury after 24 hours of exposure to a sub-lethal dose of the drug (Schyman et al., 2018). They showed that this approach could be used to predict in vivo results from gene responses in the test tube. Importantly, they found that the gene module approach could also connect the dots between rat and human in vitro data. “This research is still in the early stages, but we see great potential to use the gene module approach as a complement to in vitro high-throughput assays to predict liver and kidney injuries,” said Dr. Schyman.



References:

  • Schyman, P., Printz, R.L., Estes, S.K., Boyd, K.L., Shiota, M., and Wallqvist, A. (2018). Identification of the toxicity pathways associated with thioacetamide-induced injuries in rat liver and kidney. Front. Pharmacol. 9, 1272.
  • Te, J.A., AbdulHameed, M.D.M., and Wallqvist, A. (2016). Systems toxicology of chemically induced liver and kidney injuries: histopathology-associated gene co-expression modules. J. Appl. Toxicol. 36, 1137.


This article was published in the November 2019 issue of the TATRC Times.


Links