Methodology & Sequencing QC for 16S Metagenomics
Species-level assignments
Species-level assignments from 16S rRNA sequencing should be interpreted with caution. Species assignments beyond the top species abundance profile should be treated as provisional, owing to limitations in sequence resolution, variation among 16S gene copies and reference-database classification. For genera and species complexes known to be unresolvable by 16S, including Vibrio and other closely related aquaculture-relevant taxa, a high-abundance assignment reflects a genuinely dominant taxon, but the specific species within the clade cannot be resolved by 16S alone. Such taxa are therefore reported at the genus or species-complex level, with the best-match species given only as provisional (e.g. “Vibrio, Harveyi clade – best match V. alginolyticus“). Assignments below the reporting threshold are grouped as Vibrio sp. rather than named at species level, and any actionable pathogen assignment (e.g. V. parahaemolyticus / AHPND) requires confirmation by an independent species-specific method regardless of abundance.
Sample preparation
DNA is extracted using the MagMAX™ CORE nucleic acid purification kit with the KingFisher FLEX robot (Thermo Fisher Scientific, Australia) according to the manufacturer’s instruction.
ONT 16S sequencing
Near full-length 16S rRNA gene amplicons (targeting 16S V1-V9 variable regions) are generated using the Oxford Nanopore Technologies Microbial Amplicon Barcoding kit following manufacturer’s instructions. Individually barcoded amplicons with rapid sequencing adapters are sequenced on the Oxford Nanopore Technologies MinION/GridION flow cell and sequencing device.
Bioinformatic analysis
Base-called, demultiplexed reads are processed through Genics proprietary bioinformatics pipelines. Taxonomic classification is then performed using the NCBI 16S/18S reference database (ncbi_16s_18s).
Method sequencing QC
Every sample is scored against five checks of the sequencing run itself, and takes the worst individual result as the code reported in Table 1 of the MHA report. A sample that fails any check is excluded from the analysis.
Read count
Fewer than 1,000 reads fails. A barcode that is sparse is normally either empty or was lost during demultiplexing, and neither yields a community worth describing. Between 1,000 and 4,999 reads warns: the sample is usable, but diversity estimates drawn from it are noisy, because rare organisms may simply not have been sampled.
Read length
Near full-length 16S is about 1,500 bp, so length is a direct check on whether the intended product was sequenced. A mean outside 900-2,200 bp, or a median outside 800-2,200 bp, fails: shorter reads indicate truncation, longer ones indicate chimeric products. A mean and median differing by more than 300 bp warns, since that gap suggests a bimodal distribution — two populations of read length mixed together rather than one clean amplicon.
Base quality
Fewer than 85% of reads reaching a mean Phred quality of Q20 fails, the error rate being too high to assign reads to a species with confidence; between 85 and 90% warns. Q20 corresponds to 99% base-call accuracy.
GC content
Outside 40-70% fails. The 16S gene sits near 50-55% GC across the bacterial domain, so a result well beyond that range points to the wrong amplicon having been amplified.
