In Silico Approaches in Predictive Genetic Toxicology

In silico toxicology methods facilitate clear interpretation of the observed experimental genetic toxicity trends and provide faster, more economical, and animal-free tools for the benefit of environmental and human health in conjunction with other relevant in vitro and in vivo methods. Regulatory agencies are emphasizing the use of computational approaches to predict genetic toxicity to reduce the use of animals in toxicity testing of chemicals. Combination of in vivo (existing data), in vitro, high-throughput, and content screening data together with computational predictions might improve the predictive confidence in the genotoxicity assessment. This is an immensely difficult task, and twenty-first-century toxicology will not become animal-free overnight but is likely to use relevant and reliable computational approaches that will evolve and adapt to best use these scientific and technological advances. Expert-based, statistical QSAR models, and read-across are the most commonly used computational methodologies, and each predictive model adheres to the OECD QSAR validation principle along with an expert review system to conclude. The key steps involved in the in silico genotoxicity prediction are identification of the problem, data collection, generation of chemical descriptors, construction of the Q/SAR model, internal/external validation of the model, statistical evaluation of the models, and optimization of modeling parameters to further enhance SAR performance, updating a validated model with new chemical sets and finally reporting and documentation of the complete study. This chapter describes various in silico toxicology approaches and standardized protocols for conducting genetic toxicity predictions of chemicals and also highlights various parameters for the validation of the prediction results obtained from QSAR models.