부산대학교 도서관

Toxicogenomics has delivered predictive transcriptomic signatures for renal tubular injury, non-genotoxic carcinogenicity and numerous other pathological phenotypes. The quality and performance of transcriptomic signatures are proportional to the size of the training data set. Larger training data give less overestimation of performance in data-split-validation simulations, and therefore closer estimates to actual forward validation results. But transcriptomic models can only predict pathology for the species they are evaluated in. Better ways of translating toxic mechanisms in animal models to human biology is key to predictive toxicology. The Entelos Drug Induced Liver Injury model is a top-down, physiological model of mechanisms of liver injury in human, mouse and rat built in collaboration with the Hamner Institutes and the FDA. Modeling in Physiolab allows focus on known toxicological mechanisms and elucidation of known species differences. The initial mechanisms modeled include hepatocellular necrosis, apoptosis, cholestasis, steatosis and mitochondrial dysfunction. The use of a mechanistic dynamic model of mechanisms of liver toxicity enables simulation of effects of drugs on liver physiology, expressing the quantitative differences between species, and predictive testing of new drug candidates. [ABSTRACT FROM AUTHOR]