StackHCV: a web-based integrative machine-learning framework for large-scale identification of hepatitis C virus NS5B...

Area of Science:

• Computational chemistry and bioinformatics
• Machine learning in drug discovery
• Hepatitis C virus (HCV) research

Background:

• Identifying effective Hepatitis C Virus (HCV) NS5B polymerase inhibitors is crucial but challenging.
• Conventional experimental methods are time-consuming and resource-intensive for drug development.
• Existing computational methods often rely on single-feature approaches, limiting their scope.

Purpose of the Study:

• To develop a novel machine learning-based meta-predictor, StackHCV, for accurate and large-scale identification of HCV inhibitors.
• To improve upon existing methods by integrating diverse molecular fingerprints and machine learning algorithms.
• To provide a freely accessible web server for high-throughput screening of potential HCV drugs.

Main Methods:

• Constructed a pool of baseline models using five machine learning algorithms (k-NN, MLP, PLS, RF, SVM) with heterogeneous molecular fingerprints.
• Integrated baseline models using a stacking strategy to develop the final meta-predictor, StackHCV.
• Evaluated StackHCV performance through extensive benchmarking on training and independent test datasets.

Main Results:

• StackHCV demonstrated superior accuracy and stability compared to individual baseline models on the training dataset.
• The meta-predictor significantly outperformed existing predictors on an independent test dataset.
• A web server was developed for accessible, high-throughput identification of HCV inhibitors.

Conclusions:

• StackHCV is a powerful tool for fast and precise identification of potential drugs targeting HCV NS5B.
• The meta-predictor offers a significant advancement over single-feature approaches in HCV inhibitor discovery.
• StackHCV is expected to facilitate drug development for HCV, including applications in liver cancer therapy.