Predicting drug-target interactions from drug structure and protein sequence using novel convolutional neural...

Area of Science:

• Computational chemistry
• Bioinformatics
• Drug discovery

Background:

• Accurate identification of drug-target interactions (DTIs) is crucial for accelerating drug discovery.
• Experimental methods for DTI detection are resource-intensive and time-consuming.
• Computational approaches are vital for large-scale prediction of drug-target associations.

Purpose of the Study:

• To develop a deep learning-based method for predicting DTIs.
• To utilize only drug structures and protein sequences for DTI prediction.
• To improve the efficiency and accuracy of drug-target interaction identification.

Main Methods:

• A deep learning model was developed.
• The model was trained and evaluated using drug structures and protein sequences.
• Performance was assessed on a created dataset and a DrugBank dataset.

Main Results:

• The method achieved high accuracies (up to 92.0%) for various target families (enzymes, ion channels, GPCRs, nuclear receptors).
• On a DrugBank dataset, the model yielded an accuracy of 0.9015 and an AUC of 0.9557.
• The model demonstrated superior performance compared to state-of-the-art computational methods.

Conclusions:

• The proposed deep learning model effectively predicts drug-target interactions.
• The model extracts nuanced features, outperforming existing computational methods.
• This approach serves as a practical tool for new drug discovery and development.