Prediction of drug-target interactions via neural tangent kernel extraction feature matrix factorization model

Area of Science:

• Computational Biology
• Bioinformatics
• Machine Learning in Drug Discovery

Background:

• Drug discovery is a resource-intensive process, often taking years and significant investment.
• Predicting drug-target interactions (DTIs) is crucial for efficient drug development.
• Current machine learning methods offer potential to accelerate DTI prediction.

Purpose of the Study:

• To develop and evaluate a novel machine learning method for predicting DTIs.
• To leverage automatic feature extraction using deep learning for improved DTI prediction.
• To reduce the time and cost associated with drug development.

Main Methods:

• Utilized a neighborhood regularized logistic matrix factorization approach.
• Extracted drug and target features using a neural tangent kernel (NTK) model.
• Constructed Laplacian matrices from extracted features for matrix factorization.

Main Results:

• The proposed method demonstrated significantly superior performance across four gold standard datasets.
• Automatic feature extraction via deep learning proved competitive with manual feature selection.
• The method effectively predicted drug-target interactions, outperforming existing approaches.

Conclusions:

• The NTK-based feature extraction combined with matrix factorization offers a powerful tool for DTI prediction.
• This approach has the potential to substantially accelerate the drug discovery pipeline.
• Deep learning-based automatic feature extraction is a viable and effective strategy in computational drug discovery.