Data science approaches to pharmacogenetics

Pharmacogenetic studies rely on applied statistics to evaluate genetic data describing natural variation in response to pharmacotherapeutics such as drugs and vaccines. In the beginning, these studies were based on candidate gene approaches that specifically focused on efficacy or adverse events correlated with variants of single genes. This hypothesis driven method required the researcher to have a priori knowledge of which genes or gene sets to investigate. According to rational design, the focus of these studies has been on drug metabolizing enzymes, drug transporters, and drug targets. As technology has progressed, these studies have transitioned to hypothesis-free explorations where markers across the entire genome can be measured in large scale, population based, genome-wide association studies (GWAS). This enables identification of novel genetic biomarkers, therapeutic targets, and analysis of gene-gene interactions, which may reveal molecular mechanisms of drug activities. Ultimately, the challenge is to utilize gene-drug associations to create dosing algorithms based individual genotypes, which will guide physicians and ensure they prescribe the correct dose of the correct drug the first time eliminating trial-and-error and adverse events. We review here basic concepts and applications of data science to the genetic analysis of pharmacologic outcomes.