MFR-DTA: a multi-functional and robust model for predicting drug-target binding affinity and region

Area of Science:

• Computational biology
• Drug discovery
• Bioinformatics

Background:

• Deep learning is the standard for predicting drug-target binding affinity.
• Existing methods have limitations in sequence feature representation and biological knowledge integration for binding region prediction.

Purpose of the Study:

• To propose a novel Multi-Functional and Robust Drug-Target binding Affinity prediction (MFR-DTA) method.
• To address deficiencies in sequence element feature extraction and biological verification of predicted binding regions.

Main Methods:

• Developed BioMLP for individual biological sequence element feature extraction.
• Introduced an Elem-feature fusion block to refine features.
• Constructed a Mix-Decoder block for simultaneous drug-target interaction and binding region prediction.

Main Results:

• Evaluated MFR-DTA on two benchmarks and a new dataset, sc-PDB.
• Demonstrated superior performance over state-of-the-art methods in binding affinity prediction.
• Visualized binding sites and predicted multi-scale interaction regions, confirming accuracy.

Conclusions:

• MFR-DTA offers improved drug-target binding affinity prediction by enhancing sequence representation.
• The method provides accurate simultaneous prediction of binding regions, aiding biological verification.
• MFR-DTA shows significant advantages over existing approaches in drug discovery research.