Benchmarking Bioinformatic Tools for Amplicon-Based Sequencing of Norovirus

Area of Science:

• Environmental microbiology
• Bioinformatics
• Virology

Background:

• Accurate norovirus surveillance requires validated wet-lab and computational methods.
• No universally endorsed bioinformatics pipelines currently exist for norovirus analysis.

Purpose of the Study:

• To benchmark denoising and clustering bioinformatics pipelines for norovirus amplicon sequencing data.
• To evaluate the accuracy of open-source classifiers across different databases.
• To provide recommendations for optimal norovirus HTS data analysis.

Main Methods:

• Comparison of denoising (DADA2, Deblur, USEARCH-UNOISE3) and clustering (VSEARCH, FROGS) pipelines using simulated norovirus sequencing data.
• Evaluation of open-source classifiers (RDP, BLASTn, IDTAXA, QIIME2 naive Bayes, SINTAX) trained on custom, NoroNet, and Human calicivirus databases.
• Assessment of pipeline and classifier performance based on composition, sequence representation, and classification accuracy.

Main Results:

• VSEARCH demonstrated robust composition analysis for viral amplicons.
• Pipeline choice influenced results, with DADA2 producing more false positives and FROGS leading to underclassification.
• Classifier choice had a greater impact on classification accuracy than database choice, especially for norovirus GII.4 variants.
• All pipelines could generate Operational Taxonomic Units (OTUs) similar to expected sequences.

Conclusions:

• Recommends using the Ribosomal Database Project (RDP) classifier with VSEARCH for norovirus analysis.
• Emphasizes the critical need for maintaining up-to-date bioinformatics databases for accurate classification.
• Provides essential method standardization for analyzing norovirus amplicon HTS data, applicable to outbreak investigations.