Quick start¶

List and inspect recipes¶

bioflow recipe list                      # every recipe + stage count
bioflow recipe show prokaryote_assembly  # render the DAG without running
bioflow recipe run prokaryote_assembly --dry-run

Run a recipe¶

Recipes take their inputs as --key value options. Missing required inputs produce a clear error telling you what to pass.

Prokaryote assemblyRNA-seq DEGVariant callingPangenome (taxon-driven)

bioflow recipe run prokaryote_assembly \
  --r1 reads_R1.fastq.gz --r2 reads_R2.fastq.gz \
  --sample-id ecoli_42 --out ./out

bioflow recipe run rnaseq_deg \
  --sample-sheet samples.csv --transcriptome ref.fa \
  --out ./out

bioflow recipe run germline_variants \
  --r1 sample_R1.fq.gz --r2 sample_R2.fq.gz \
  --reference genome.fa --snpeff-db Escherichia_coli_k12 \
  --sample-id sample01 --out ./out

bioflow recipe run pangenome --taxon Dickeya --max 13 --out ./out

Recipes use input-hash caching automatically — a second run with the same inputs returns in seconds. Stages that declare it retry with bumped resources on failure (e.g. SPAdes → 2× RAM on OOM).

Reference databases¶

Some recipes need external databases (Kraken2 DB, BUSCO lineages, etc.).

bioflow db list                              # what's available
bioflow db fetch kraken2_standard_8gb --dest /refs
bioflow db verify kraken2_standard_8gb --dest /refs

Programmatic use (Tier-A SDK)¶

from bioflow import stage, pipeline, set_workspace
from bioflow.recipes import get

set_workspace("./out")
result = get("prokaryote_assembly")(
    r1="R1.fq.gz", r2="R2.fq.gz", out_dir="./out", sample_id="demo",
)
print(result.ok, result.out_dir)

See examples/recipes_quickstart.py for the call signature of every recipe.