We found that the combined use of Pollux and Fiona gave the best results when error-correcting Ion Torrent PGM viral data. Of the algorithms tested, Pollux showed a better overall performance but tended to over-correct ‘genuine’ substitution variants, whereas Fiona proved to be better at distinguishing these variants from sequencing errors. Tests using both real and simulated data showed that the algorithms differed in their abilities to detect and correct errors and that the error rate and sequencing depth significantly affected the performance. Deletion errors were clearly present at the ends of homopolymer runs. We examined 19 quality-trimmed PGM datasets for the HBV reverse transcriptase (RT) region and found a total error rate of 0.48% ± 0.12%.
A full systematic assessment of the effectiveness of various error correction algorithms in PGM viral datasets (e.g., hepatitis B virus (HBV)) has not been performed. Ion Torrent Personal Genome Machine (PGM) technology is a mid-length read, low-cost and high-speed next-generation sequencing platform with a relatively high insertion and deletion (indel) error rate.
