Lineage classification and marker-driven genotyping — from assemblies or raw reads.

📖 Read the documentation

Tutorials, every command option, input formats, output columns and benchmarks.

Paula Ruiz-Rodriguez¹ and Mireia Coscolla¹
_{1. Institute for Integrative Systems Biology, I²SysBio, University of Valencia-CSIC, Valencia, Spain}

---

What is pathotypr?

pathotypr is a Rust toolkit that classifies microbial genomes into lineages and genotypes them against user-defined marker panels. It works with both assembled genomes (FASTA) and raw sequencing reads (FASTQ), runs on a single laptop, and ships with a native desktop GUI.

Command	What it does	Input	Guide
train	Build a Random Forest classifier from labeled genomes	FASTA	docs
predict	Assign lineages using a trained model	FASTA + model	docs
classify	Call known SNP markers in assemblies	FASTA + markers	docs
split-fastq	Alignment-free genotyping from reads	FASTQ + markers	docs
match	Find the closest reference genome	FASTQ + references	docs

Note

Nothing is hard-coded to one organism: the marker panel you supply defines what is typed. In practice pathotypr has only been validated on the M. tuberculosis complex, so treat other organisms as exploratory and check them against a truth set you trust.

Install

Command line

conda create -n pathotypr -c bioconda pathotypr
conda activate pathotypr
pathotypr --help

Desktop app: download an installer for macOS, Linux or Windows. No compiler needed.

Building from source, system dependencies and the first-launch notes for unsigned apps are in the installation guide.

Quick start

# Genotype an assembly against a marker panel
pathotypr classify -m markers.tsv -r reference.fasta -i sample.fasta -o results

# Genotype straight from reads
pathotypr split-fastq -m markers.tsv -r reference.fasta \
  -i reads_R1.fastq.gz -i reads_R2.fastq.gz -o genotype

# Train a model, then apply it
pathotypr train   -i labeled_genomes.fasta -o model.pathotypr.zst
pathotypr predict -i query.fasta -m model.pathotypr.zst -o predictions.tsv

Add --excel to any command to also write .xlsx. The getting started tutorial walks through a full MTBC run, from install to reading the output.

Ready-to-use MTBC data

Marker panels and a pre-trained model for the M. tuberculosis complex are published on Zenodo, DOI 10.5281/zenodo.19210044:

File	Contents
pathotypr_lineage_markers_v1.0.0.tsv	3,707 lineage SNPs (L1–L10, A1–A4)
pathotypr_dr_markers_v1.0.0.tsv	102,213 DR mutations (WHO catalogue v2, 2023)
pathotypr_rf_model_v1.0.0.pathotypr	Pre-trained Random Forest (k=31, 100 trees)

Download links and usage are in the installation guide.

Documentation

Everything lives at pathogenomics-lab.github.io/pathotypr:

Getting started	End-to-end MTBC tutorial
Input formats	What every file must look like, per command
Marker format	Curating your own panel
Output files	Every column of every file
Desktop GUI	The app, and building it
Algorithms	How each module works
Benchmarks	Speed, memory and tool comparison
FAQ	Common problems

Citation

If you use pathotypr, please cite:

Ruiz-Rodriguez P, Coscollá M. Pathotypr: harmonised MTBC lineage assignment and resistance-associated variant detection for genomic surveillance. bioRxiv (2026). doi: 10.64898/2026.03.24.714002

BibTeX, RIS and APA entries, plus the software DOI, are on the citation page.

License

GNU Affero General Public License v3.0

---

✨ Contributors

pathotypr is developed with ❤️ by:

Paula Ruiz-Rodriguez 💻 🔬 🤔 🔣 🎨 🔧

Mireia Coscolla 🔍 🤔 🧑‍🏫 🔬 📓

This project follows the all-contributors specification (emoji key).