Pathway-based drug repositioning using causal inference.

Area of Science:

• Computational biology
• Pharmacology
• Systems biology

Background:

• Recent in vivo studies highlight drug repositioning potential via causality inference.
• Existing methods often use a simplified 'one drug-one target-one disease' model.
• There is a need for systematic in silico approaches to identify novel therapeutic applications for drugs.

Purpose of the Study:

• To develop and present an in silico method, CauseNet, for building a causal network between drugs and diseases.
• To systematically identify new therapeutic uses for existing drugs through computational analysis.
• To leverage causality inference for advancing drug repositioning strategies.

Main Methods:

• Constructed a causal network (CauseNet) integrating drugs, targets, pathways, and genes.
• Utilized expert-curated knowledge bases for comprehensive network information.
• Employed statistical learning to estimate causal chain transition likelihoods based on known drug-disease associations.

Main Results:

• Achieved high performance in cross-validation with an AUC of 0.859.
• Top-ranked predictions demonstrated significant enrichment in scientific literature and clinical trials.
• Identified several potential drug repurposing candidates for Crohn's Disease.

Conclusions:

• Successfully developed a computational method for discovering new drug uses via causal inference.
• The CauseNet approach enables hypothesis generation for drug repositioning using accessible biological data.
• This layered network analysis facilitates the identification of novel therapeutic applications for existing medications.