LiBRe: A Ligand-Aware Sequence-Based Binding Residue Prediction Model for Virtual Screening.

Area of Science:

• Computational biology
• Drug discovery
• Bioinformatics

Background:

• Identifying protein-ligand binding residues is crucial for understanding molecular recognition and developing new therapeutics.
• Existing sequence-based deep learning models often neglect ligand information, despite binding being ligand-specific.

Purpose of the Study:

• To develop a novel ligand-aware sequence-based deep learning model for predicting protein-ligand binding residues.
• To improve the accuracy of binding residue prediction by integrating both protein and ligand data.

Main Methods:

• Proposed a sequence-based deep learning model incorporating residue-level protein sequence information and ligand features.
• Evaluated the model's performance against existing sequence-based and structure-based methods.

Main Results:

• The ligand-aware model significantly improved the prediction of ligand-binding residues compared to current baselines.
• Binding pockets identified using the model's predictions exhibited stronger and more stable binding affinity.

Conclusions:

• The proposed model demonstrates superior performance in predicting protein-ligand binding residues.
• This approach holds significant potential for advancing virtual screening and accelerating drug discovery pipelines.