Genetically supported drug target prioritization for rare diseases

Area of Science:

• Genomics
• Pharmacology
• Computational Biology

Background:

• Rare and uncommon diseases often lack targeted therapies due to limited research.
• Existing drug target prioritization methods struggle with the complexity of rare disease genetics.

Purpose of the Study:

• To develop and validate a machine-learning framework, RareGPS, for prioritizing drug targets in rare diseases.
• To improve the prediction of drug indications and clinical trial progression for rare disease therapeutics.

Main Methods:

• RareGPS integrates 11 sources of genetic, clinical, and experimental evidence.
• It employs an allelic-series model using the full distribution of genetic associations across allele-frequency bins.
• The framework was validated using prescriptome analyses in two million patients and the AMELIE literature evaluation tool.

Main Results:

• RareGPS outperforms existing resources in predicting drug indications and clinical trial progression across 161 phenotypes.
• Top 1% prioritized targets showed a 58-fold higher likelihood of advancing from nonindicated to Phase IV trials.
• Targets also demonstrated an 8-fold higher likelihood of advancing from Phase I to Phase IV trials compared to the middle 50%.

Conclusions:

• RareGPS provides a robust computational approach for identifying high-potential drug targets for rare diseases.
• The framework's predictions for 3,021,965 gene-phenotype pairs can accelerate rare disease drug discovery.
• RareGPS enhances the efficiency of drug development pipelines for unmet medical needs in rare conditions.