graphLambda: Fusion Graph Neural Networks for Binding Affinity Prediction.

Area of Science:

• Computational chemistry
• cheminformatics
• bioinformatics

Background:

• Accurate prediction of protein-ligand binding affinity is essential for drug discovery.
• Deep learning scoring functions show promise for predicting binding constants.
• Graph neural networks (GNNs) offer advanced capabilities for molecular modeling.

Purpose of the Study:

• To introduce graphLambda, a novel GNN-based model for enhanced protein-ligand binding affinity prediction.
• To improve the accuracy and robustness of binding affinity prediction for computer-aided drug discovery (CADD).

Main Methods:

• Utilized graph convolutional, attention, and isomorphism blocks within the GNN architecture.
• Developed a novel deep learning model named graphLambda.
• Evaluated model performance on established benchmarks like CASF16 and CSAR HiQ NRC.

Main Results:

• graphLambda demonstrated superior predictive performance on CASF16 and CSAR HiQ NRC benchmarks.
• The model showed robustness across various train-validation set partitioning strategies.
• Achieved enhanced predictive capabilities through specialized graph neural network blocks.

Conclusions:

• graphLambda represents a significant advancement in GNN-based binding affinity prediction.
• The model holds potential for more effective CADD methodologies.
• Highlights the growing importance of GNNs in computational drug discovery.