FRnet-DTI: Deep convolutional neural network for drug-target interaction prediction.

Area of Science:

• Pharmacology
• Computational Biology
• Drug Discovery

Background:

• Drug-target interaction (DTI) prediction is crucial for developing new therapeutics.
• Accurate DTI prediction accelerates the drug design process.
• Existing methods face challenges in feature representation and predictive accuracy.

Purpose of the Study:

• To introduce FRnet-DTI, a novel deep learning framework for DTI prediction.
• To enhance feature representation using an auto-encoder and classification using CNNs.
• To validate the efficacy of FRnet-DTI on established benchmark datasets.

Main Methods:

• FRnet-DTI employs an auto-encoder (FRnet-Encode) for feature extraction, generating 4096 features per instance.
• A convolutional neural network (FRnet-Predict) is utilized for classifying drug-target interactions based on these features.
• The model was evaluated on four widely-used gold standard DTI datasets.

Main Results:

• FRnet-DTI significantly outperformed state-of-the-art methods on three out of four datasets.
• Performance improvements were observed in both area under the Receiver Operating Characteristic curve (auROC) and area under the Precision-Recall curve (auPR).
• The study identified twenty novel potential drug-target pairs with high predicted interaction scores.

Conclusions:

• FRnet-DTI offers a robust and effective approach for drug-target interaction prediction.
• The proposed deep learning framework demonstrates superior performance compared to existing methods.
• The identified potential drug-target pairs warrant further experimental validation for therapeutic development.