Skip to main content

phylotypy

PyPI version Python 3.11+ License: MIT

A Naive Bayesian Classifier for 16S rRNA gene sequences, inspired by the phylotypr R package by Riffomonas. Designed for classifying amplicon sequence variants (ASVs) from DADA2, QIIME2, or raw FASTA files against a reference database of 16S rRNA sequences. The RDP training data is provided here in the data directory located at the github repository. But Silva and others can be used.

Thanks to Riffomonas for the inspiration — check out the videos on his YouTube channel.


Performance

The full RDP reference database takes ~7.5 seconds and the full Silva reference database (genus level) takes ~19 seconds on a 2020 Apple Intel MacBook Pro with 16Gb of RAM. Newer systems should see a substantial increase in performance.


How to Install

Using pip:

pip install phylotypy

Using uv (recommended — how to install uv):

uv pip install phylotypy

Note: Intel Mac (x86_64) users are limited to numba 0.62.1, which is pinned in this package. Apple Silicon (M-series) users are not affected.


Training Data

Download the RDP reference training set and an example dataset before classifying:

File Description
rdp_16S_v19.dada2.fasta RDP trainset19072023, DADA2 format
dna_moving_pictures.fasta Example dataset (Moving Pictures study)
The training data fasta descriptions should contain a taxonomy header only. This RDP training data is
genus level, but phylotypy can accept species level training data.
"Kingdom", "Phylum", "Class", "Order", "Family", "Genus", "Species"

The taxon string in the fasta description should follow the semicolon-separated format like this:

>Bacteria;Pseudomonadota;Gammaproteobacteria;Enterobacterales;Enterobacteriaceae;Citrobacter
TAGAGTTTGATCCATGGCTCAGATTGAACGCTGGCGGCAGGCCTAACAC.....

Reference Database: Taxonomic Levels

Every reference sequence needs a taxonomy string in its FASTA header (see the format above), and all headers in the file must have the same number of ;-delimited levels, e.g. Kingdom;Phylum;Class;Order;Family;Genus is 6 levels. This matters because make_classifier() and classify_sequences() compare taxonomy strings assuming a fixed depth; a reference set with mixed depths will fail, either right away or partway through classification.

1. Headers already share the same depth

If your reference fasta is already uniform (true of the RDP training set used above, at 6 levels/genus), just load and build normally:

from phylotypy import read_fasta, classifier

rdp = read_fasta.read_taxa_fasta("rdp_16S_v19.dada2.fasta", is_ref=True)
database = classifier.make_classifier(rdp, verbose=True)

Pass is_ref=True to read_taxa_fasta() to check this up front: it counts the semicolon-delimited levels in every header, and raises a ValueError naming the file and the depths it found if they're not all the same. So a bad reference file is caught immediately, not partway through building the classifier or later during classification. Leave is_ref at its default (False) when reading sequences you're going to classify, since those don't carry a fixed-depth taxonomy string.

2. Fixing ragged headers

"Ragged" headers means the taxonomy strings in the reference fasta don't all reach the same depth. Some sequences are annotated down to Genus, others only down to Class or Order:

>Bacteria;Firmicutes;Bacilli;Lactobacillales;Lactobacillaceae;Lactobacillus
TAGAGTTTGATCCTGGCTCAG...
>Bacteria;Firmicutes;Bacilli
TAGAGTTTGATCCTGGCTCAG...

This is common with broader databases like Silva. Trying to build a classifier from ragged reference data (or reading it with is_ref=True) raises:

ValueError: ref_db contains taxonomy strings with inconsistent numbers of taxonomic levels:
id
3      1204
6     88213
All 'id' entries must have the same number of ';'-delimited levels before building a
classifier, or classify_sequences() will fail later on.
Options:
  - Fix/pad the taxonomy strings in ref_db so every entry has the same depth.
  - Call make_classifier(ref_db, filter_db=True) to filter ref_db down to a single
    consistent depth (see training_data.filter_train_set).
  - Pass n_levels=<int> together with filter_db=True to control which depth is kept.

Fix it with training_data.filter_train_set(), which drops sequences that don't match a target depth (n_levels) and strips out common noise terms (Eukaryota, metagenome, Candidatus, etc.. See training_data.DEFAULT_NOISE_TERMS):

from phylotypy import read_fasta, training_data, classifier

# is_ref left at False here — the file is ragged, so the check would raise
silva = read_fasta.read_taxa_fasta("silva_nr99.fasta.gz")

filtered = training_data.filter_train_set(silva, n_levels=6, verbose=True)
database = classifier.make_classifier(filtered, verbose=True)

With verbose=True, filter_train_set() reports how many sequences it dropped:

Removed 12,458 sequences with taxonomic depth != 6 (kept 158,970)

Or skip the manual filter step and let make_classifier() do it for you:

database = classifier.make_classifier(silva, filter_db=True, verbose=True)

This runs filter_train_set() internally, auto-detecting n_levels from the majority depth in your data if you don't pass one, and prints the same before/after counts when verbose=True.


Quick Start

1. Load training data and sequences to classify

from phylotypy import classifier, results, read_fasta

rdp = read_fasta.read_taxa_fasta("rdp_16S_v19.dada2.fasta")
moving_pics = read_fasta.read_taxa_fasta("dna_moving_pictures.fasta")

2. Train the classifier

# Accepts fasta or csv/tsv with 'id' and 'sequence' column names
database = classifier.make_classifier(rdp, verbose=True)

# If the reference has ragged (inconsistent) taxonomic levels, use filter_db=True
# to remove the mismatched records — see "Reference Database: Taxonomic Levels" above
database = classifier.make_classifier(rdp, filter_db=True, verbose=True)

3. Classify sequences

classified = classifier.classify_sequences(moving_pics, database)

4. Format and export results

classified = results.summarize_predictions(classified)
print(classified.columns)

Output:

Index(['id', 'sequence', 'classification', 'Kingdom', 'Phylum', 'Class',
       'Order', 'Family', 'Genus', 'observed', 'lineage'],
      dtype='object')
classified.to_csv("classified_results.csv")

Complete Code Block

from phylotypy import classifier, results, read_fasta

rdp = read_fasta.read_taxa_fasta("rdp_16S_v19.dada2.fasta")
moving_pics = read_fasta.read_taxa_fasta("dna_moving_pictures.fasta")

database = classifier.make_classifier(rdp)

classified = classifier.classify_sequences(moving_pics, database)
classified = results.summarize_predictions(classified)
print(classified.head())

classified.to_csv("classified_results.csv")

Example Classification Output

Taxonomic levels (Domain → Genus) are semicolon-separated. Numbers in parentheses represent bootstrap confidence scores. The default confidence threshold is 80%.

Bacteria(100);Pseudomonadota(99);Alphaproteobacteria(99);Rhodospirillales(99);Acetobacteraceae(99);Roseomonas(83)

Bacteria(99);Bacteroidota(97);Bacteroidia(93);Bacteroidales(93);Bacteroidales_unclassified(93);Bacteroidales_unclassified(93)

Bacteria(100);Bacteroidota(100);Bacteroidia(100);Bacteroidales(100);Bacteroidaceae(100);Bacteroides(100)

Working with Your Own Data

phylotypy works with FASTA files from DADA2, QIIME2, or any standard pipeline. See read_fasta.py for utilities to load and convert sequence data into the required format.

A complete walkthrough is available in vignette.py.


Requirements

Dependencies are installed automatically via pip. See pyproject.toml for the full list.


Citation

If you use phylotypy in your research, please cite:

  • Wang, Q., Garrity, G.M., Tiedje, J.M., Cole, J.R. (2007) Naive Bayesian Classifier for Rapid Assignment of rRNA Sequences into the New Bacterial Taxonomy. Applied and Environmental Microbiology, 73(16), 5261–5267.
  • Schloss PD.2025.phylotypr: an R package for classifying DNA sequences. Microbiol Resour Announc14:e01144-24.https://doi.org/10.1128/mra.01144-24
  • Saltikov, C. (2024) phylotypy: Python implementation of a Naive Bayesian 16S rRNA classifier. https://github.com/csaltikov/phylotypy

Release files for phylotypy 0.5.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for phylotypy 0.5.0
File Size Uploaded
phylotypy-0.5.0.tar.gz 307.7 kB Details

Built distributions (wheels)

Table of built distributions (wheels) for phylotypy 0.5.0
File
phylotypy-0.5.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.manylinux_2_28_x86_64.whl CPython 3.13 CPython 3.13 Linux glibc 2.17+ x86-64, Linux glibc 2.28+ x86-64 Details
phylotypy-0.5.0-cp313-cp313-macosx_10_13_universal2.whl CPython 3.13 CPython 3.13 macOS 10.13+ universal2 (ARM64, x86-64) Details
phylotypy-0.5.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.manylinux_2_28_x86_64.whl CPython 3.12 CPython 3.12 Linux glibc 2.17+ x86-64, Linux glibc 2.28+ x86-64 Details
phylotypy-0.5.0-cp312-cp312-macosx_10_13_universal2.whl CPython 3.12 CPython 3.12 macOS 10.13+ universal2 (ARM64, x86-64) Details
phylotypy-0.5.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.manylinux_2_28_x86_64.whl CPython 3.11 CPython 3.11 Linux glibc 2.17+ x86-64, Linux glibc 2.28+ x86-64 Details
phylotypy-0.5.0-cp311-cp311-macosx_10_9_universal2.whl CPython 3.11 CPython 3.11 macOS 10.9+ universal2 (ARM64, x86-64) Details

Total release size: 5.1 MB

Release files / phylotypy-0.5.0.tar.gz

Download URL phylotypy-0.5.0.tar.gz
Size 307.7 kB
Tags Source
SHA-256 checksum
How to use checksums
afbfaed32cacd6021e0816aa5957f8b2ec7aeab2ec9c9a3cea7117d684e7484e
BLAKE2b-256 checksum
How to use checksums
226c9282544257eaa7239e6e12294b420ee16b49960688030c70f2fa6d31842c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 31, 2026.

Transparency log

Release files / phylotypy-0.5.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.manylinux_2_28_x86_64.whl

Download URL phylotypy-0.5.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.manylinux_2_28_x86_64.whl
Size 1.1 MB
Tags CPython 3.13 Linux glibc 2.17+ x86-64 Linux glibc 2.28+ x86-64
SHA-256 checksum
How to use checksums
813d9cf95e6bec6b3ba38a6af33912802942ddf7b80d41e93c4597db5270c877
BLAKE2b-256 checksum
How to use checksums
4fefd3389e0eb85e6f4f07fac105c7a6e1bc0c8dcd1b16e21e17f3161466c9d3
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 31, 2026.

Transparency log

Release files / phylotypy-0.5.0-cp313-cp313-macosx_10_13_universal2.whl

Download URL phylotypy-0.5.0-cp313-cp313-macosx_10_13_universal2.whl
Size 547.3 kB
Tags CPython 3.13 macOS 10.13+ universal2 (ARM64, x86-64)
SHA-256 checksum
How to use checksums
9e53da4e2a883e8aeed9fb3f75a67722cf70ab2be660bcc70699e41ae591e46e
BLAKE2b-256 checksum
How to use checksums
ae63bd8b0f88e6e638cf5609a8348fec1504ba29eb6396920221cdbba9dd0d42
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 31, 2026.

Transparency log

Release files / phylotypy-0.5.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.manylinux_2_28_x86_64.whl

Download URL phylotypy-0.5.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.manylinux_2_28_x86_64.whl
Size 1.1 MB
Tags CPython 3.12 Linux glibc 2.17+ x86-64 Linux glibc 2.28+ x86-64
SHA-256 checksum
How to use checksums
8cce6895e00f8f271e2dfaed23d4a81d87e2570f3bb05e61c75965c67e5676a5
BLAKE2b-256 checksum
How to use checksums
4ddb039d4e74884f3e8e946755800df066127924f4a5681142219b15e6ae965d
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 31, 2026.

Transparency log

Release files / phylotypy-0.5.0-cp312-cp312-macosx_10_13_universal2.whl

Download URL phylotypy-0.5.0-cp312-cp312-macosx_10_13_universal2.whl
Size 548.0 kB
Tags CPython 3.12 macOS 10.13+ universal2 (ARM64, x86-64)
SHA-256 checksum
How to use checksums
fac71980d345757679cc5af5a262cba06b658837b72b43ba22f239924c6aa36e
BLAKE2b-256 checksum
How to use checksums
ba0b5a43764229f6ad8cd35e9ba4f72472aa2953af7dd52f039ad4f116b3f9aa
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 31, 2026.

Transparency log

Release files / phylotypy-0.5.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.manylinux_2_28_x86_64.whl

Download URL phylotypy-0.5.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.manylinux_2_28_x86_64.whl
Size 1.0 MB
Tags CPython 3.11 Linux glibc 2.17+ x86-64 Linux glibc 2.28+ x86-64
SHA-256 checksum
How to use checksums
f9c21c2dc90fe899881d6115d8f999add470b31b0a47e83290c2c36f65ecc82c
BLAKE2b-256 checksum
How to use checksums
d417584eaf9177a9f7aa9372c416977ec7481947337b8f3697ec9eb6bb9cdce9
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 31, 2026.

Transparency log

Release files / phylotypy-0.5.0-cp311-cp311-macosx_10_9_universal2.whl

Download URL phylotypy-0.5.0-cp311-cp311-macosx_10_9_universal2.whl
Size 545.1 kB
Tags CPython 3.11 macOS 10.9+ universal2 (ARM64, x86-64)
SHA-256 checksum
How to use checksums
6b8b8ea599275d622f107c7d64bc46bfdf071ee155ef4996ddb8cfbe014896b6
BLAKE2b-256 checksum
How to use checksums
12b2c9fe656f4909a8624ce585383fc96fbc397c4b02f890ac83ee52ea29a461
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Aug 31, 2026.

Transparency log

Release history Release notifications | RSS feed

0.9.1

7 release files

0.9.0

7 release files

0.7.5

7 release files

0.6.0

7 release files

This release

0.5.0 This release

7 release files

0.4.0

7 release files

0.3.4

7 release files

0.3.2

10 release files

0.3.1

1 release file

0.3.0

4 release files

0.2.7

4 release files

0.2.6

4 release files

0.2.5

4 release files

0.2.4

3 release files

0.2.3

3 release files

0.2.2

3 release files

0.2.0

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page