CLI reference¶

Global flags (before the subcommand): -V/--version (-J for JSON), -v/--verbose, -q/--quiet, -h/--help. Every command accepts -t/--threads (default 4).

run¶

Full pipeline from FASTQ: fastp QC → bowtie2 host removal (optional) → profiling → reports.

# single sample
gmeta run sample_R1.fq.gz sample_R2.fq.gz \
  -s SAMPLE-001 -m kraken2 \
  -D /data/db/kraken2 --host-index /data/db/host/grch38 \
  -f json -f html -f xlsx -f pdf \
  -O gmeta-out/SAMPLE-001 -t 16

# batch: manifest (sample_id<TAB>r1[<TAB>r2] per line) with 4 concurrent samples
gmeta run -S samples.tsv -m kraken2 -D /data/db/kraken2 -J 4 -O gmeta-out

# batch: positional reads grouped by mate markers (A_R1 A_R2 B_R1 B_R2 → A, B)
gmeta run A_R1.fq A_R2.fq B_R1.fq B_R2.fq -m kmcp -D /data/db

Batch runs write per-sample directories (<outdir>/<sample>/…) and emit one JSON line per sample on stdout (order preserved even with -J); a failed sample yields {"sample": …, "error": …} without aborting the batch.

Flag	Meaning
`-s/--sample-id`	sample identifier (single-sample runs; batches derive ids from filenames)
`-S/--samples`	batch manifest: `sample_id<TAB>reads [<TAB>reads2]` per line
`-J/--jobs`	concurrent samples in batch runs (default 1)
`-m/--method`	kraken2 (default) / sylph / centrifuge / kaiju / metaphlan / kmcp
`-D/--db-dir`	profiler database root; default `$GMETA_DB_DIR`
`--host-index`	bowtie2 host index prefix; omit to skip host removal
`-O/--outdir`	output directory (single: gmeta-out/; batch: gmeta-out//*)
`-f/--format`	report formats, repeatable: json html xlsx pdf (default json+html)
`--sample-type` `--platform` `--lab` `--operator` `--sequencing-date`	metadata recorded in the report

Artifacts land in <outdir>/pipeline/ (kreport.txt, bracken.txt, centrifuge.tsv, kaiju.*.tsv, profile.tsv …) so reports can cite them.

report¶

Build a report from existing results — no re-run.

gmeta report -s SAMPLE-001 --kreport k.txt --bracken b.txt --fastp-json f.json
gmeta report -s SAMPLE-001 --metaphlan profile.txt -o SAMPLE-1.report.html

Flag	Meaning
`--kreport` / `--bracken`	kraken2 outputs (either alone works)
`--fastp-json`	QC metrics for the report
`--sylph-profile`	sylph profile TSV
`--centrifuge`	centrifuge native report
`--kaiju` / `--kaiju-genus`	kaiju2table species / genus tables
`--metaphlan`	MetaPhlAn profile
`--kmcp`	KMCP profile TSV
`-o/--output`	output file; format from extension (.json/.html/.xlsx/.pdf), default JSON to stdout
`--top` / `--min-fraction`	row filtering (default 50 / 0.0)

net build¶

识别网 submission workbook — see 识别网 export.

db¶

gmeta db list                    # database locations + state
gmeta db fetch [TARGETS]...      # generate download scripts (not executed)

Fetch targets: kraken2 centrifuge kaiju metaphlan kmcp sylph taxdump host, or all. Options: -O/--out-dir (default db-fetch), --db-root (install root baked into the scripts). See Databases.