Fusion-Based Deep Learning Architecture for Detecting Drug-Target Binding Affinity Using Target and Drug Sequence and...

Area of Science:

• Computational chemistry
• Drug discovery
• Bioinformatics

Background:

• Accurate drug-target binding affinity prediction is crucial for efficient drug discovery.
• Existing computational methods often focus on either target sequence or structure, neglecting integrated information.
• Wet laboratory experiments for binding affinity are costly and time-consuming.

Purpose of the Study:

• To develop a novel computational model, CGraphDTA, for predicting drug-target binding affinity.
• To integrate both target sequence and molecular structure information for improved prediction accuracy.
• To accelerate the drug discovery process through enhanced computational predictions.

Main Methods:

• Developed CGraphDTA, a fusion protocol using multiscale convolutional neural networks (CNNs) and graph neural networks (GNNs).
• Utilized CNNs to extract sequence-based features from drugs and targets.
• Employed GNNs to extract structure-based features from drug and target molecular graphs.

Main Results:

• CGraphDTA demonstrated superior performance compared to state-of-the-art methods on test datasets.
• Ablation studies confirmed the effectiveness of the integrated sequence and structure approach.
• Biological interpretation and drug selectivity evaluations validated CGraphDTA's utility.

Conclusions:

• CGraphDTA effectively predicts drug-target binding affinity by leveraging both sequence and structural data.
• The model offers a significant advancement over existing methods, accelerating drug discovery pipelines.
• CGraphDTA serves as a valuable computational tool for identifying potential drug candidates.