Prediction of drug-target interaction based on protein features using undersampling and feature selection techniques...

Area of Science:

• Bioinformatics
• Computational Chemistry
• Drug Discovery

Background:

• Accurate drug-target interaction (DTI) identification is vital for drug discovery but traditional methods are costly and time-consuming.
• Existing computational methods for DTI prediction still leave many interactions undiscovered.

Purpose of the Study:

• To develop a novel computational model, pdti-EssB, for accurate identification of drug-target interactions.
• To leverage protein sequence and drug molecular structure information for enhanced DTI prediction.

Main Methods:

• Drug molecules were represented as molecular substructure fingerprints.
• Protein sequences were described using evolutionary, sequence, and structural information.
• Data balancing techniques and a novel feature eliminator were employed with an XGBoost model.

Main Results:

• The pdti-EssB model demonstrated superior performance in predicting DTIs compared to recent methods.
• The model successfully identified new potential drug-target interaction samples based on prediction scores.
• Performance was validated using four benchmark DTI datasets and five-fold cross-validation, with auROC as the evaluation metric.

Conclusions:

• The pdti-EssB model offers a highly effective computational approach for DTI identification.
• This method can significantly expedite the drug discovery process by predicting novel interactions.
• The pdti-EssB webserver is publicly accessible for broader research application.