Novel target identification towards drug repurposing based on biological activity profiles

Area of Science:

• Computational biology
• Pharmacology
• Genomics

Background:

• Rare diseases impact over 30 million people, with significant unmet needs for effective treatments.
• Drug repurposing offers a viable strategy to accelerate therapeutic development by leveraging existing compounds.
• Identifying novel gene targets and compound relationships is crucial for successful drug repurposing.

Purpose of the Study:

• To develop and validate predictive machine learning models for identifying potential drug repurposing candidates.
• To uncover latent relationships between gene targets and chemical compounds for rare disease therapies.
• To streamline the drug discovery pipeline for rare diseases through computational approaches.

Main Methods:

• Machine learning algorithms including Support Vector Classifier, K-Nearest Neighbors, Random Forest, and Extreme Gradient Boosting were employed.
• Models were trained on comprehensive biological activity profile data for 143 gene targets and over 6000 compounds.
• Predictions were validated using public experimental datasets and case studies.

Main Results:

• Developed predictive models with high accuracy (greater than 0.75) for identifying gene-target and compound relationships.
• Successfully predicted potential therapeutic targets for drug repurposing.
• Validated model predictions through independent datasets and specific case study evaluations.

Conclusions:

• Computational modeling and machine learning significantly enhance the efficiency of drug repurposing for rare diseases.
• This approach provides a robust framework for identifying novel therapeutic interventions.
• The study facilitates the discovery of more effective treatments, addressing critical needs in rare disease populations.