Exploring the relationship between drug side-effects and therapeutic indications

Area of Science:

• Pharmacology and Cheminformatics
• Computational Drug Discovery

Background:

• Predicting therapeutic indications and drug side-effects is crucial but challenging in drug development.
• Existing methods often rely solely on chemical structures or protein targets.

Purpose of the Study:

• To compare the predictive power of indication information for side-effect prediction and vice versa, against traditional structure- and target-based models.
• To identify highly correlated disease-side-effect pairs for drug repositioning and adverse reaction prediction.

Main Methods:

• Utilized 10-fold cross-validation to evaluate prediction models.
• Compared models incorporating indication-side-effect relationships with those using only chemical structures or protein targets.
• Extracted statistically significant disease-side-effect pairs from known drug-disease and drug-side-effect relationships.

Main Results:

• Drug side-effects and therapeutic indications were found to be the most predictive features for each other.
• Identified 6,706 highly correlated disease-side-effect pairs.
• Demonstrated the utility of these pairs for suggesting drug repositioning and adverse reaction watch lists.

Conclusions:

• Integrating indication and side-effect data significantly enhances prediction accuracy in drug discovery.
• The identified disease-side-effect relationships offer valuable insights for both therapeutic indication discovery and safety profiling.
• The study provides a data-driven approach to uncover novel drug applications and potential risks.