A lightweight tool for one purpose only: to unify the outputs of different taxonomic classifiers. Currently supports MMseqs2, Centrifuge v1.0.4, Kraken2 v2.1.3, Metabuli v1.0.1, MetaMaps v.633d2e. The output files of these tools are converted to contigs predictions format.

The tab-delimitered output file has two columns: sequence identifiers (the 1st column, contigs) and taxonomic labels on all levels from domain to species concatenated with ";" (predictions).

IMPORTANT: if you come from the Taxometer README page (https://github.com/RasmussenLab/vamb/blob/taxometer_release/README_Taxometer.md), run the command with the --mmseqs-format flag. The output file will have a different format but the same information. Taxometer, when run from the release branch (taxometer_release), uses this format. It is also possible to run Taxometer from the newest release of VAMB, in which case you don't need the flag. This compatibility issue will be fixed in the future releases of the VAMB library.

Suggestions and contributions are most welcome.

1. Clone this repo and install the package from the source (releasing pip package WIP).

git clone [email protected]:RasmussenLab/taxconverter.git
cd taxconverter
pip install -e .

2. Unzip the two files from data/lineage.zip (38.3 MB): ncbi_lineage.csv (246.2 MB) and metabuli_lineage.csv (58.1 MB), and place them to the data/ folder.

To convert Centrifuge, Kraken2, MetaMaps and MMSeqs2 outputs, provide one file with the taxonomy annotation results:

taxconverter centrifuge -i centrifuge_annotations.tsv -o result.tsv

taxconverter kraken2 -i kraken2_annotations.tsv -o result.tsv

taxconverter metamaps -i metamaps_annotations.tsv -o result.tsv

taxconverter mmseqs2 -i metamaps_annotations.tsv -o result.tsv

To convert a Metabuli output, provide two files with _classifications.tsv and _report.tsv postfixes:

taxconverter metabuli -c metabuli_classifications.tsv -r metabuli_report.tsv -o result.tsv

For more help, run taxconverter -h, taxconverter metabuli -h, taxconverter centrifuge -h, taxconverter kraken2 -h, taxconverter metamaps -h, taxconverter mmseqs2 -h

This package is made to complement the Taxometer tool for refining taxonomic annotations from any classifier using contigs k-mers and co-abundances (Article Youtube).

Other links:

• MMseqs2 Article Github
• Metabuli Article Github
• Centrifuge Article Github
• Kraken2 Article Github
• MetaMaps Article Github