Dereplicate and subsample FASTQ files, writing one JSON file per sample.
dada2-rs sample reads/*.fastq.gz -o sampled/ --nbases 1e8 --gzipProcesses input FASTQ files in order (or shuffled with --randomize),
dereplicating each and writing a JSON file to --output-dir. Processing stops
once the cumulative base count reaches --nbases.
Each output file uses the same format as derep and can be passed
directly to errors-from-sample. Pre-sampling is the
way to spread error learning across many samples rather than letting one deep
sample dominate — see the caveat on
learn-errors.
<INPUT>... — FASTQ files (.fastq, .fastq.gz, .fq, .fq.gz).
--phred-offset — 33 for Sanger / Illumina 1.8+, 64 for Illumina 1.3–1.7.
--nbases (default 1e8) — stop after accumulating at least this many total
bases across input files. Files are taken whole.
--randomize — process input files in random order instead of the supplied
order. This shuffles sample order only; reads are never subsampled within a
file.
--seed — RNG seed for reproducible --randomize.
--threads (default 1) — threads for dereplication.
--output-dir / -o — directory for per-sample JSON files, created if
absent.
--pretty — pretty-print the JSON; the default is compact (~34% smaller).
--gzip — write {sample}.json.gz, read back transparently downstream.
--verbose — progress to stderr.