Figure 1: Overview and workflow used in this study.

(a) Studies examining side effect pathogenesis focus primarily on drug pharmacokinetics, involving drug transport and clearance, and drug binding in terms of on and off target-binding events. This study examines potential pathogenic mechanisms related to transcriptional changes downstream of clearance and binding events. (b) Drug-treated gene expression profiles from the Connectivity Map database are analysed in the context of the metabolic network reconstruction Recon 1 using constraint-based modelling to identify drug-induced pathway expression changes. Drug-induced metabolic pathway expression changes are analysed in terms of drug side effects from the Side Effect Resource (SIDER) using a feature selection genetic algorithm to determine metabolic pathway perturbations conserved in particular side effects, termed DISLoDGED pathways. (c) A new database, the Metabolism Disease Database (MDDB), was generated by manual curation of literature to establish links between altered metabolic pathway function and pathologies, and this database was used to analyse DISLoDGED metabolic pathways. (d) Five candidate causal mechanisms for metabolic changes in side effect pathogenesis (listed in the MDDB panel) are assessed in a large-scale manner by comparing these in vitro perturbations to clinical data linking particular metabolic pathways to disease.