ViVaMBC: estimating viral sequence variation in complex populations from illumina deep-sequencing data using...

Area of Science:

• Genomics
• Bioinformatics
• Virology

Background:

• Deep-sequencing technologies enable comprehensive analysis of sequence variations in populations.
• Sequencing errors can hinder accurate detection of low-frequency mutations.
• Illumina sequencing quality scores and secondary base calls offer potential for error correction.

Purpose of the Study:

• To introduce ViVaMBC, a novel model-based clustering method for identifying and quantifying viral variants.
• To leverage quality scores and second-best base calls for enhanced variant detection.
• To enable codon-level variant calling for direct biological interpretation, particularly for antiviral drug responses.

Main Methods:

• Developed ViVaMBC, a virus variant model-based clustering approach.
• Utilized Illumina sequencing quality scores and second-best base calls.
• Optimized variant calling at the codon level for biological relevance.

Main Results:

• ViVaMBC accurately estimates variant frequencies down to 0.5% with unbiased results at 25,000x coverage.
• Demonstrated superior sensitivity and specificity compared to V-Phaser2, ShoRAH, and LoFreq for variants >0.4%.
• Observed increased false positives below 0.4%, potentially due to sample/library preparation errors.

Conclusions:

• ViVaMBC is the first method for direct, codon-level viral variant calling.
• Quality score-based error modeling is the primary strength, significantly reducing sequencing errors.
• Second-best base calls offered marginal sensitivity gains, not justifying computational overhead; PCR errors become a limiting factor.