SSUplex: fast, both-strand extraction and origin-sorting of small-subunit rRNA for environmental DNA metabarcoding

Motivation Long-read sequencing has made full-length small-subunit ribosomal RNA metabarcoding routine, but broadly conserved primers also amplify organellar and cross-domain sequences, which distort profiles when sent to a target-domain classifier. Results SSUplex detects small-subunit ribosomal RNA on both strands with conserved-region profile hidden Markov models, assigns each read to one of five origins (bacteria, archaea, eukaryota, mitochondria, chloroplast) and extracts the gene region. It reimplements Metaxa2’s extraction step but ranks origins by summed rather than mean per-region score. On 5,000 ZymoBIOMICS Nanopore reads, this raised accuracy from 54.9% for Metaxa2’s extraction step to 95.9%, against 99.96% for Metaxa2’s full pipeline with similarity-search classification. On reads simulated from sources outside the Metaxa2 database, SSUplex was 93 to 98% accurate for bacteria-dominated and plant-root-like compositions (Metaxa2’s full pipeline: 96 to 99%), with mitochondria the weakest origin. In a rice-root 16S sample, about 40% of reads were organellar. SSUplex ran 3.2× faster than Metaxa2’s extraction step from 200,000 reads upward, with about 45% less peak memory, and 65 to 134× faster than Metaxa2’s full pipeline. Availability and implementation SSUplex is implemented in Rust and freely available at https://github.com/ayobi/ssuplex and from Bioconda; version 0.2.0 is archived at https://doi.org/10.5281/zenodo.23246210.