papa2

Python-first amplicon denoising — byte-identical to R's DADA2

papa2 is a complete Python port of the DADA2 amplicon denoising pipeline. All 37 R functions have Python equivalents, producing byte-identical results (20/20 parity tests pass) with no R dependency.

Full documentation: https://rec3141.github.io/papa2

---

Key Features

• Byte-identical to R dada2 — same ASVs, same error models, same quality filtering, validated against R on every function; filtered FASTQ output is byte-identical to R's, sequence tables match R in both counts and column order, and seeded taxonomy bootstraps reproduce an R session's set.seed(N) stream
• Complete port — all 37 user-facing R functions: filter_and_trim, derep_fastq, learn_errors, dada, merge_pairs, make_sequence_table, remove_bimera_denovo, assign_taxonomy, and more
• No R dependency — standalone Python package with a compiled C/C++ core (libpapa2.so)
• Fast — vectorized numpy filtering with isal-accelerated gzip, OpenMP in the C core, and sample-level worker pools; cores are split automatically between samples and within-sample threads (DADA2_WORKERS / DADA2_OMP_THREADS to override)
• Interactive visualisation — Sankey diagrams for read tracking, error rate plots, quality profiles
• Container-ready — Docker and Apptainer (HPC) images on GHCR

---

Performance

Measured on a 32-core machine against multithreaded R dada2 1.40, real 2x151 MiSeq data (details and parity guarantees):

Workload	R dada2	papa2	Speedup
6 samples end-to-end	284 s	156 s	1.8x
120 samples end-to-end	1019 s	512 s	2.0x
filterAndTrim (single-thread)	69 s	10 s	7x
assignTaxonomy, SILVA 138.1 x 2,000 ASVs	71 s	15 s	4.7x
assignTaxonomy, SILVA 138.1 x 20,697 ASVs	~35 min	50 s	~40x

---

Quick Install

Container (recommended for HPC)

# Docker
docker pull ghcr.io/rec3141/papa2:latest
docker run -v $(pwd):/data ghcr.io/rec3141/papa2 python3 my_script.py

# Apptainer (HPC)
apptainer pull papa2.sif docker://ghcr.io/rec3141/papa2:latest
apptainer exec papa2.sif python3 my_script.py

From source

git clone https://github.com/rec3141/papa2.git
cd papa2
conda env create -f environment.yml
conda activate papa2-dev
make libpapa2.so
pip install -e .

---

Usage

import papa2

# Forward and reverse FASTQ files
fwd_files = ["sample1_R1.fastq.gz", "sample2_R1.fastq.gz"]
rev_files = ["sample1_R2.fastq.gz", "sample2_R2.fastq.gz"]

# 1. Filter and trim
papa2.filter_and_trim(
    fwd_files, filt_fwd,
    rev=rev_files, filt_rev=filt_rev,
    trunc_len=(240, 200), max_ee=(2, 2), rm_phix=True,
)

# 2. Learn error rates
errF = papa2.learn_errors(filt_fwd)
errR = papa2.learn_errors(filt_rev)

# 3. Dereplicate, denoise, merge
derepFs = [papa2.derep_fastq(f) for f in filt_fwd]
derepRs = [papa2.derep_fastq(f) for f in filt_rev]
dadaFs = papa2.dada(derepFs, err=errF)
dadaRs = papa2.dada(derepRs, err=errR)
mergers = [papa2.merge_pairs(dF, drF, dR, drR)
           for dF, drF, dR, drR in zip(dadaFs, derepFs, dadaRs, derepRs)]

# 4. Sequence table and chimera removal
seqtab = papa2.make_sequence_table(mergers)
seqtab_nochim = papa2.remove_bimera_denovo(seqtab)

# 5. Taxonomy
taxa = papa2.assign_taxonomy(seqtab_nochim["seqs"], "silva_nr99_v138.1_train_set.fa.gz")

# 6. Visualise
track = papa2.track_reads(dereps=derepFs, dadas=dadaFs, mergers=mergers,
                          seqtab=seqtab, seqtab_nochim=seqtab_nochim, taxa=taxa)
papa2.plot_sankey(track, output="read_tracking.html")

See the Quickstart, Tutorial, and Big Data Workflow for complete walkthroughs.

---

Project Layout

papa2/           Python package
  filter.py        filter_and_trim, fastq_filter, fastq_paired_filter
  io.py            derep_fastq
  dada.py          dada, learn_errors
  error.py         loess_errfun, pacbio_errfun, make_binned_qual_errfun
  taxonomy.py      assign_taxonomy
  paired.py        merge_pairs
  chimera.py       remove_bimera_denovo
  utils.py         sequence tables, plotting, QC, taxonomy, export
  _cdada.py        ctypes bindings to libpapa2.so
src/             Standalone C/C++ core
tests/           Parity tests against R dada2
docs/            MkDocs documentation source

---

Provenance

papa2 was extracted from rec3141/dada2@gpu-python, which forked from upstream DADA2 at commit 72da7700b (2026-02-13).

---

Citation

If you use papa2 in published research, please cite the original DADA2 paper:

Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP (2016). DADA2: High-resolution sample inference from Illumina amplicon data. Nature Methods, 13, 581–583. https://doi.org/10.1038/nmeth.3869

---

License

LGPL-3.0 — see LICENSE.