CSatDTA: Prediction of Drug-Target Binding Affinity Using Convolution Model with Self-Attention.

Area of Science:

• Computational chemistry
• Pharmacology
• Bioinformatics

Background:

• Drug discovery is vital for identifying novel treatments, involving chemistry, pharmacology, and biology.
• Accurate prediction of drug-target affinity (DTA) is critical in early drug development stages.

Purpose of the Study:

• To introduce a novel model, CSatDTA, for predicting drug-target affinity.
• To enhance DTA prediction by integrating convolution with self-attention mechanisms.

Main Methods:

• CSatDTA applies convolution-based self-attention to molecular drug and target sequences.
• The model leverages self-attention for capturing long-range interactions, addressing limitations of traditional convolutional neural networks (CNNs).

Main Results:

• CSatDTA demonstrates superior performance compared to existing sequence-based and other DTA prediction methods.
• The model exhibits significant retention abilities in predicting drug-target interactions.

Conclusions:

• CSatDTA offers an effective approach for predicting drug-target affinity.
• The integration of self-attention mechanisms enhances the predictive power for drug discovery.