Machine Learning and Deep Learning Applications to Evaluate Mutagenicity

Mutagenicity evaluation for chemical agents is an important aspect of risk assessment. In vitro assays, such as bacterial mutation assay and mammalian cell gene mutation assay, have been widely used to evaluate mutagenicity, and the use of in vivo studies is much less frequent. Due to the high cost and time consuming nature of the experimental methods, computational approaches, including machine learning (ML) and deep learning (DL) methods based on chemical structure, for predicting mutagenicity are recognized as useful alternatives. The use of in silico methods is rapidly increasing, and the methods are now becoming more broadly accepted by regulatory agencies for certain purposes. This chapter outlines the state of the art of applying ML and DL methods to predict mutagenicity and their role in chemical risk assessment.