Enrichment analysis of phenotypic data for drug repurposing in rare diseases

Abstract

Drug-induced Behavioral Signature Analysis (DBSA), is a machine learning (ML) method for in silico screening of compounds, inspired by analytical methods quantifying gene enrichment in genomic analyses. When applied to behavioral data it can identify drugs that can potentially reverse in vivo behavioral symptoms in animal models of human disease and suggest new hypotheses for drug discovery and repurposing. We present a proof-of-concept study aiming to assess Drug-induced Behavioral Signature Analysis (DBSA) as a systematic approach for drug discovery for rare disorders. We applied Drug-induced Behavioral Signature Analysis to high-content behavioral data obtained with SmartCube®, an automated in vivo phenotyping platform. The therapeutic potential of several dozen approved drugs was assessed for phenotypic reversal of the behavioral profile of a Huntington’s Disease (HD) murine model, the Q175 heterozygous knock-in mice. The in silico Drug-induced Behavioral Signature Analysis predictions were enriched for drugs known to be effective in the symptomatic treatment of Huntington’s Disease, including bupropion, modafinil, methylphenidate, and several SSRIs, as well as the atypical antidepressant tianeptine. To validate the method, we tested acute and chronic effects of tianeptine (20 mg/kg, i. p.) in vivo, using Q175 mice and wild type controls. In both experiments, tianeptine significantly rescued the behavioral phenotype assessed with the SmartCube® platform. Our target-agnostic method thus showed promise for identification of symptomatic relief treatments for rare disorders, providing an alternative method for hypothesis generation and drug discovery for disorders with huge disease burden and unmet medical needs.