EBOLApred

EBOLApred predicts small molecules with potential inhibitory activity against Ebola virus glycoprotein (GP) and matrix protein VP40 using machine learning models trained on EBOV cell entry inhibitor data from PubChem.

Key Features:

• Targets: Focuses on Ebola virus glycoprotein (GP) and matrix protein VP40 as molecular targets.
• Input data: Trained on EBOV cell entry inhibitor compounds sourced from the PubChem database.
• Prediction scope: Classifies small molecules for potential inhibition of GP and VP40 activities.
• Algorithms: Implements five machine learning algorithms: random forest (RF), support vector machine (SVM), naïve Bayes (NB), k-nearest neighbor (kNN), and logistic regression (LR).
• Validation: Models were evaluated using 10-fold cross-validation.
• Performance: Random forest achieved 89% accuracy, F1 score 0.9, and ROC AUC 0.95, while logistic regression and SVM achieved overall accuracies of 84% and 86%, respectively.

Scientific Applications:

• Anti-Ebola drug discovery: Computational identification of candidate inhibitors targeting GP and VP40 for Ebola virus research.
• Prioritization for experimental validation: Selection of promising small-molecule candidates for subsequent experimental testing against Ebola virus proteins.

Methodology:

Models were trained on EBOV cell entry inhibitor data from PubChem using random forest, support vector machine, naïve Bayes, k-nearest neighbor, and logistic regression and were evaluated by 10-fold cross-validation.