Interdisciplinary computational approaches that combine statistics, computer science, medicine, chemoinformatics, and biology are becoming highly valuable for drug1 discovery and development. Data mining and machine learning methods are being more commonly used to properly analyze the emerging high volumes of structured and unstructured biomedical and biological data from several sources including hospitals, laboratories, pharmaceutical companies, and even social media. These data may include sequencing and gene expression, drug molecular structures, protein and drug interaction networks, clinical trial and electronic patient records, patient behavior and selfreporting data in social media, regulatory monitoring data, and biomedical literature. Data mining methods can be used in several stages of drug discovery and development to achieve different goals. Figure 1.1 summarizes the drug development and FDA2 approval process diagram. Most new compounds fail during this approval process in clinical trials or cause adverse side effects. The cost of successful novel chemistry-based drug development often reaches millions of dollars, and the time to introduce the drug to market often comes close to a decade [1]. The high failure rate of drugs during this process, make the trial phases known as the “valley of death”[2]. Similar to many other domains, pharmaceutical data mining algorithms aim to limit the search space and provide recommendations to domain experts for hypothesis generation and further analysis and experiments. One way to categorize data mining and machine learning approaches is based on their application to pre-marketing and post-marketing stages. In pre-marketing stage, data mining methods focus on discovery activities, including but not limited to, finding signals that indicate relations between drugs and targets, drugs and drugs, genes and diseases, protein and diseases, and finding bio-markers. In this stage potential interactions that could cause therapeutic or adverse effects are studied. Most of the chemical compounds under study at this stage have not been through clinical trails, and the in silico experiments serve as a basis for further explorations for them. In postmarketing stage an important application of data analytics is in finding indications of adverse side effects for approved drugs. These algorithms provide a list of potential drug side effect associations that can be used for further studies.

In this chapter we provide a brief overview of some data analytics applications in this domain, and mainly focus on two major tasks from each stage. We first summarize some of the main methods for drug-target interaction prediction that is highly important during the pre-marketing stage. We then provide an overview of pharmacovigilance (or drug safety surveillance) which is an important focus in the post-marketing stage.