Large-Scale Off-Target Identification Using Fast and Accurate Dual Regularized One-Class Collaborative Filtering and...

Area of Science:

• Computational chemistry
• Pharmacology
• Bioinformatics

Background:

• Drug discovery relies on target-based screening, but off-target interactions can cause side effects or therapeutic benefits.
• Identifying these off-target interactions is vital for developing safe and effective drugs and exploring drug repurposing opportunities.
• Existing computational methods for predicting drug-target interactions are often inaccurate or too slow for large-scale analysis.

Purpose of the Study:

• To develop a fast and accurate computational method for predicting genome-wide drug off-target interactions.
• To evaluate the proposed method's performance across different gene families on a proteome scale.
• To identify potential drug repurposing candidates for novel anti-cancer therapies.

Main Methods:

• Developed REMAP, a computational method based on a dual regularized one-class collaborative filtering algorithm.
• Explored chemical space, protein space, and their interactions on a large scale.
• Validated REMAP using a reliable, extensive, and cross-gene family benchmark dataset.

Main Results:

• REMAP demonstrated superior accuracy compared to state-of-the-art methods in predicting off-target interactions across gene families.
• The method is highly scalable, capable of screening 200,000 chemicals against 20,000 proteins in under 2 hours.
• Predicted seven FDA-approved drugs for repurposing as anti-cancer therapies, with six supported by existing experimental evidence.

Conclusions:

• REMAP is a valuable and efficient in silico tool for drug target identification, drug repurposing, phenotypic screening, and side effect prediction.
• The method facilitates large-scale exploration of chemical and protein spaces for drug discovery.
• Identified promising drug repurposing candidates for anti-cancer treatment, warranting further investigation.