Anti-Ebola

Anti-Ebola predicts inhibitory potency of small molecules against Ebola virus using regression-based machine learning models trained on a curated DrugRepV dataset of 305 anti-Ebola compounds with IC50 values.

Key Features:

• Data Source and Preparation: Curated dataset of 305 unique anti-Ebola compounds from the DrugRepV database annotated with IC50 values.
• Molecular Descriptor Extraction: Extraction of molecular descriptors to support a quantitative structure-activity relationship (QSAR) framework.
• Machine Learning Models: Regression-based models built using support vector machines (SVM), random forests, and artificial neural networks (ANN).
• Model Performance and Validation: Tenfold cross-validation with Pearson's correlation coefficients ranging from 0.83 to 0.98 on training and testing datasets (T^274) and robustness assessed via William’s plot.

Scientific Applications:

• Inhibitor prediction: Predicts potential Ebola virus inhibitors to prioritize compounds for experimental validation.
• Virtual screening: Enables screening of large compound libraries to identify candidates with high predicted inhibitory potency.
• QSAR and model development: Supports quantitative structure-activity relationship studies and development and validation of predictive models for antiviral research.

Methodology:

Curated 305-compound dataset from DrugRepV with IC50 annotations; extraction of molecular descriptors for QSAR; regression modeling using support vector machines (SVM), random forests, and artificial neural networks (ANN); tenfold cross-validation and evaluation by Pearson's correlation coefficients (0.83–0.98) on training and testing datasets (T^274); and robustness assessment using William’s plot.