HCVpred

HCVpred predicts the bioactivity of compounds targeting the hepatitis C virus (HCV) NS5B polymerase to support the discovery of NS5B inhibitors.

Key Features:

• Classification Structure-Activity Relationship (CSAR) model: A CSAR model developed from a dataset of 578 non-redundant compounds focuses on identifying substructures associated with anti-HCV activity.
• Fingerprint descriptors and Random Forest: Twelve fingerprint descriptors characterize NS5B inhibitors and predictive models are constructed using the Random Forest algorithm across 100 independent data splits.
• Modelability and performance metrics: Dataset modelability is reported with a MODI index of 0.88 (threshold 0.65), and predictive performance metrics (accuracy, sensitivity, specificity, Matthews correlation coefficient) are reported as statistically robust with values exceeding 0.8 and 0.7, respectively.
• Descriptor analysis: Analysis highlights the importance of aromatic rings and alkyl side chains for effective NS5B inhibition.

Scientific Applications:

• NS5B inhibitor discovery: Predicting active versus inactive compounds against HCV NS5B to prioritize and guide design of potential therapeutic agents for hepatitis C.

Methodology:

Input molecules in SMILES notation are converted to twelve fingerprint descriptors, evaluated by a classification SAR model using Random Forest trained across 100 independent data splits, with modelability assessed by the MODI index (0.88).