LGBMDF: A cascade forest framework with LightGBM for predicting drug-target interactions.

Area of Science:

• Computational chemistry
• Bioinformatics
• Machine learning in drug discovery

Background:

• Drug-target interactions (DTIs) are crucial for drug development.
• Traditional laboratory methods for DTI determination are time-consuming and costly.
• Machine learning offers a faster, more cost-effective approach to predict DTIs.

Purpose of the Study:

• To develop a highly accurate and computationally efficient model for predicting drug-target interactions (DTIs).
• To improve upon existing deep forest-based DTI prediction methods by replacing XGBoost with LightGBM.

Main Methods:

• A novel cascade forest model, LGBMDF, was developed using LightGBM and ExtraTrees estimators.
• The model's performance was evaluated using 5-fold cross-validation on a standard DTI dataset.
• Key performance metrics including Sn, Sp, MCC, AUC, and AUPR were compared against state-of-the-art methods.

Main Results:

• LGBMDF demonstrated superior prediction performance compared to existing state-of-the-art methods.
• The proposed LGBMDF model exhibited a significantly faster calculation speed.
• The experimental determination of the estimator group as three LightGBMs and three ExtraTrees optimized performance.

Conclusions:

• LGBMDF presents a promising advancement in DTI prediction, offering enhanced accuracy and efficiency.
• The model has the potential to accelerate the drug discovery and development pipeline.
• Replacing XGBoost with LightGBM in cascade forests is an effective strategy for improving DTI prediction models.