In silico drug repositioning using omics data: the potential and pitfalls

With the recognition of the heterogeneity within complex diseases, such as cancer, there is an accompanying understanding of the need for a stratified approach to treatment. Patients with different underlying biologies originating at the genomic, epigenetic, or transcriptomics levels may present with similar phenotypes at diagnosis. The same treatment may thus result in different outcomes. Using the wealth of public information that is available, particularly from high-throughput experiments, regarding the behavior of approved drugs may facilitate the discovery of novel treatments for subgroups of patients. In silico approaches to drug repositioning have been developed over the past 15 years with a view to enabling this process, with a focus on mapping compounds to patient phenotypes and uncovering novel mechanisms of action. An understanding of the core structure and design of each of these tools, possible applications, and how different inputs can influence results is essential in order that users can maximize the potential of such in silico analyses. This in turn will accelerate the preclinical stage of the biomarker translational pipeline, often perceived as a key bottleneck.