Comparison of error correction algorithms for Ion Torrent PGM data: application to hepatitis B virus

Area of Science:

• Genomics
• Bioinformatics
• Virology

Background:

• Ion Torrent Personal Genome Machine (PGM) is a next-generation sequencing platform known for speed and cost-effectiveness.
• PGM technology exhibits a notable insertion and deletion (indel) error rate, particularly in viral genomic data.
• A comprehensive evaluation of error correction algorithms for PGM-generated viral datasets, such as hepatitis B virus (HBV), is lacking.

Purpose of the Study:

• To systematically assess the performance of various error correction algorithms on Ion Torrent PGM sequencing data for HBV.
• To identify the most effective algorithms for correcting indel errors in PGM viral datasets.
• To determine the impact of error rate and sequencing depth on algorithm performance.

Main Methods:

• Analysis of 19 quality-trimmed PGM datasets targeting the HBV reverse transcriptase (RT) region.
• Evaluation of multiple error correction algorithms using both real and simulated sequencing data.
• Assessment of algorithm performance based on error detection, correction accuracy, and handling of genuine variants.

Main Results:

• The overall error rate in the examined PGM HBV datasets was 0.48% ± 0.12%, with deletion errors prevalent at homopolymer run ends.
• Algorithms demonstrated variable efficacy in error detection and correction, influenced by error rate and sequencing depth.
• Pollux tended to over-correct genuine substitution variants, while Fiona excelled at differentiating them from sequencing errors.

Conclusions:

• The combined application of Pollux and Fiona algorithms yielded optimal results for error correction in Ion Torrent PGM viral data.
• Effective error correction is crucial for accurate variant calling and downstream analysis of PGM-generated viral sequences.
• This study provides valuable insights for optimizing bioinformatics pipelines for PGM sequencing data, particularly in viral genomics.