dada2-rs <subcommand> --help prints a one-line summary for every flag,
grouped into categories. These pages carry the detail that does not fit on one
line: what a parameter actually does, when to change it, and what the evidence
behind its default is.
Flags are grouped under a fixed set of headings, used consistently across every subcommand:
| Heading | What it holds |
|---|---|
| Input | Input files, sample naming, Phred offset |
| Error model | The error model file, fitting function, and its knobs |
| Denoising | The DADA2 algorithm parameters (R's setDadaOpt() surface) |
| Pseudo-pooling | Prior selection between the two dada-pseudo rounds |
| Alignment | Needleman-Wunsch scoring, band radius, aligner backend |
| Screening | The pre-alignment k-mer screen that decides which pairs are aligned |
| Trimming | Truncation and end-trimming of reads |
| Filtering | Read- or sequence-level discard rules |
| Primers | Primer sequences and matching tolerance |
| Merging | Paired-end overlap and concatenation rules |
| Chimera | Bimera and trimera detection parameters |
| Classification | Taxonomic assignment thresholds |
| Metrics | Optional per-read metrics (summary, summary-merge) |
| Pooling regime | Which population is scored, and how it is sampled (kdist-calibrate) |
| Evaluation | Truth-set classification thresholds (reference-eval) |
| Performance | Threads and concurrency; never affects results |
| Diagnostics | Extra output for QC and debugging; never affects ASVs |
| Experimental | Unstable flags. Off by default, and not drop-in replacements |
Two headings are worth reading as a pair. Denoising parameters define the clusters; Screening parameters only decide which pairs are worth aligning in the first place. A screen flag can change runtime by orders of magnitude and can, if set too aggressively, cost you real ASVs — but it is not part of the algorithm's definition of an ASV.
Anything under Performance or Diagnostics is safe to change without affecting the ASV table.
Quality assessment
summary— per-position quality metrics from a FASTQ filesummary-merge— union per-sample summaries into a run report
Read preparation
filter-and-trim— filter and trim one sampleremove-primers— primer detection and trimmingderep— dereplicate a FASTQ filesample— dereplicate and subsample, one JSON per sample
Error models
learn-errors— fit an error model from FASTQ or JSONerrors-from-sample— fit from pre-sampled JSON
Denoising
dada— per-sample (Rpool=FALSE)dada-pooled— full pooling (Rpool=TRUE)dada-pseudo— pseudo-pooling (Rpool="pseudo")
Tables and chimeras
merge-pairs— merge denoised read pairsmake-sequence-table— sample x sequence tableremove-bimera-denovo— bimera removalchimera-diagnostics— trimera screen
Taxonomy
assign-taxonomy— Naive Bayes k-mer classifierassign-species— exact-match species assignment
Converters
Calibration and evaluation
kdist-calibrate— calibrate the k-mer screenreference-eval— evaluate ASVs against a truth set
- Parameters:
setDadaOpt()parity — the R-equivalency table - Tuning for your data
- Findings — the measurements behind the defaults