Chapter 15
Machine Learning-Based Modeling of Drug Toxicity
Jing Lu, Dong Lu, Zunyun Fu, Mingyue Zheng, and Xiaomin Luo
Abstract
Toxicity is an important reason for the failure of drug research and development (R&D). The traditional
experimental testings for chemical toxicity profile are costly and time-consuming. Therefore, it is attractive
to develop the effective and accurate alternatives, such as in silico prediction models. In this review, we
discuss the practical use of some prediction models on three toxicity end points, including acute toxicity,
carcinogenicity, and inhibition of the human ether-a-go-go-related gene ion channel (hERG). Special
emphasis is put on the machine learning methods for developing in silico models, and their advantages
and weaknesses are discussed. We conclude that machine learning methods are valuable for helping the
process of designing new candidates with low toxicity in drug R&D studies. In the future, much still needs
to be done to understand more completely the biological mechanisms for toxicity and to develop more
accurate prediction models to screen compounds.
KeywordsMachine learning method, In silico model, Acute toxicity, Carcinogenicity, hERG1 Introduction
Toxicity of drugs and candidates is an important issue for drug
research and development (R&D), resulting in the increased attri-
tion and cost, late-stage failures, and marked withdrawals. CMR
International data demonstrated that in the period between 2006
and 2010, toxicity accounted for 22% and 54% of drug R&D fail-
ures in clinical development and preclinical development, respec-
tively [1]. Moreover, Lasser et al. reported that of 548 drugs
approved by FDA in the period 1975–2000, 10.2% had at least
one black-box warning and 2.9% were withdrawn [2]. Thus, it is
necessary to put an increased emphasis on the evaluation of toxicity
profile of compounds in the early stage of drug R&D.
Nowadays, there are substantial toxicological data in many phar-
maceutical companies and commercial databases. For example,
almost 60,000 compounds had been screened by using in vitro
hERG inhibition assay at Pfizer in 2005 [3]. The molecular clinical
safety intelligence system developed by GlaxoSmithKline includes
pharmacological, metabolic, and toxicological information ofTao Huang (ed.),Computational Systems Biology: Methods and Protocols, Methods in Molecular Biology, vol. 1754,
https://doi.org/10.1007/978-1-4939-7717-8_15,©Springer Science+Business Media, LLC, part of Springer Nature 2018
247