debar

debar corrects insertion and deletion (indel) errors in cytochrome c oxidase I (COI-5P) DNA barcode and metabarcode sequences using a profile Hidden Markov Model and the Viterbi algorithm to improve high-throughput sequencing (HTS)–based biodiversity assessments.

Key Features:

• Profile Hidden Markov Model (PHMM): Employs a PHMM tailored to the conserved structure of the COI region to identify and localize indel errors.
• Viterbi Algorithm: Applies the Viterbi algorithm to determine the most probable path through the PHMM and adjust sequences to correct indels.
• Integration with aphid: Relies on functions from the R package aphid for constructing the PHMM and executing the Viterbi algorithm.
• R package implementation: Implemented as an R package for programmatic sequence processing and denoising of COI-5P data.
• Performance metrics: Identified 95% of artificially introduced indels in silico, reduced errors by 75% for Sequel data and 94% for Ion Torrent S5 data, and reported a false correction rate below 0.1%.

Scientific Applications:

• Species discovery and biodiversity assessments: Improves accuracy of COI-based species identification and biodiversity estimates by reducing indel-induced artefacts in barcode and metabarcode data.
• High-throughput sequencing data analysis: Applied to datasets from platforms such as PacBio Sequel and Ion Torrent S5 to reduce platform-specific indel errors, with reported reductions of 75% and 94%, respectively.

Methodology:

Compares input sequences against a COI-tailored PHMM, simulates indels, applies the Viterbi algorithm to infer the most probable alignment path, and uses aphid functions to construct the PHMM and execute the Viterbi analysis.