COMPUTING THERAPY FOR PRECISION MEDICINE: COLLABORATIVE FILTERING INTEGRATES AND PREDICTS MULTI-ENTITY INTERACTIONS.

Area of Science:

• Biomedical Informatics
• Computational Biology
• Systems Biology

Background:

• Biomedicine generates vast data, yet knowledge integration across studies remains a challenge.
• Discovering connections between chemicals, genes, and diseases is crucial for understanding biological systems.

Purpose of the Study:

• To develop and validate a novel approach for integrating disparate biomedical data.
• To predict interactions among biological entities using collaborative filtering.
• To explore the potential for drug repurposing and advancing precision medicine.

Main Methods:

• Utilized a Collaborative Filtering approach based on non-negative matrix factorization (NMF).
• Applied NMF to interaction matrices from the Comparative Toxicogenomics Database (CTD): Chemical-Disease, Chemical-Gene, and Disease-Gene.
• Validated predictions using 10-fold cross-validation and external datasets like STRING.

Main Results:

• Successfully recovered known Chemical-Disease, Chemical-Gene, and Disease-Gene networks.
• Demonstrated the ability to predict interaction matrices from each other.
• Integrated CTD and STRING data to significantly improve Chemical-Gene interaction predictions.
• Showcased predictive power on time-stamped data, forecasting future information.

Conclusions:

• Collaborative filtering effectively integrates information across diverse biological entities.
• The method provides a foundation for generating drug repurposing hypotheses.
• This approach represents a step towards achieving precision medicine goals.