DGDTA: dynamic graph attention network for predicting drug-target binding affinity.

Area of Science:

• Computational chemistry
• Bioinformatics
• Machine learning in drug discovery

Background:

• Accurate drug-target binding affinity (DTA) is crucial for drug discovery and repositioning.
• Existing DTA prediction methods require further analysis of protein and drug features.
• Deep learning presents a promising avenue for improving DTA prediction.

Purpose of the Study:

• To develop a novel deep learning model for accurate drug-target binding affinity prediction.
• To enhance the analysis of drug and protein features for DTA prediction.
• To improve upon existing methods in the field of computational drug discovery.

Main Methods:

• Proposed Dynamic Graph DTA (DGDTA) model utilizing a dynamic graph attention network and a bidirectional long short-term memory (Bi-LSTM) network.
• Input includes drug compounds represented by Simplified Molecular Input Line Entry System (SMILES) and protein amino acid sequences.
• Drugs are modeled as graphs, with dynamic attention scores highlighting important atoms and edges; Bi-LSTM extracts contextual protein sequence features.

Main Results:

• DGDTA effectively predicts DTA by integrating drug and protein feature vectors through a fully connected layer.
• The model leverages dynamic attention for drug feature extraction and Bi-LSTM for protein sequence analysis.
• Source code is publicly available on GitHub.

Conclusions:

• DGDTA demonstrates superior accuracy in predicting drug-target binding affinity compared to existing methods.
• The proposed deep learning approach offers a more effective way to predict DTA.
• This advancement contributes to more efficient drug discovery and repositioning strategies.