DLSSAffinity: protein-ligand binding affinity prediction via a deep learning model

Area of Science:

• Computational chemistry
• Structural biology
• Bioinformatics

Background:

• Protein-ligand binding affinity prediction is crucial for computer-aided drug discovery.
• Current methods often rely on limited 3D structures or protein sequences, posing a challenge for accurate predictions.

Purpose of the Study:

• To develop a novel deep learning approach, DLSSAffinity, for accurate protein-ligand binding affinity prediction.
• To leverage both local structural and global sequence information for improved predictive performance.

Main Methods:

• DLSSAffinity utilizes pocket-ligand structural pairs for local interaction prediction.
• It incorporates full-length protein sequences and ligand SMILES for global interaction prediction.
• The model was evaluated on the PDBbind benchmark dataset.

Main Results:

• DLSSAffinity achieved a Pearson's R of 0.79, RMSE of 1.40, and SD of 1.35 on the test set.
• The model demonstrated superior performance compared to existing state-of-the-art deep learning methods for binding affinity prediction.

Conclusions:

• Combining global sequence and local structure information significantly improves the accuracy of protein-ligand binding affinity prediction.
• DLSSAffinity represents a promising advancement in computational drug discovery.