brucetyper

In silico Bruce-ladder, Suis-ladder, AMOS PCR and an RB51 junction PCR for Brucella, run end to end from assembled genomes.

For every FASTA file the primer pairs of each requested assay are located directly in the nucleotide sequence, the resulting amplicon sizes are read the way a gel would be read, and a genotype is called. No external PCR tool, no alignment step, no database.

Quick start

pip install brucetyper

# Every assay, on a folder of assemblies
brucetyper -i genomes/ -o results/

# One assay, on a single genome
brucetyper -i GCF_000054005.1.fna.gz -o results/ --amos

Or from a clone, where python bruce_ladder.py takes the same arguments as brucetyper:

git clone https://github.com/duceppemo/brucetyper
cd brucetyper
pip install biopython
python bruce_ladder.py -i genomes/ -o results/

Input is a FASTA file or a folder of them, gzipped or not, one sample per file. Assemblies, not reads. The sample name is the file name with its extension stripped. With no assay flag, all four are run.

Three kinds of report are written to --outdir: amplicons.tsv (every product found), calls.tsv (one row per sample and assay) and one report_<assay>.tsv per assay run (sample and call, two columns).

$ cat results/calls.tsv
sample  assay  call          rough  bands                     sizes                 extra_bands
RB51    bruce  RB51          yes    152;450-451;590;774;2526  2526;774;590;450;152
RB51    amos   RB51_or_2308         178;387;494               494;387;178
RB51    rb51   RB51                 919                       1300;919              1300
2308    bruce  abortus       no     152;450-451;590;774;1682  1682;774;590;450;152
2308    rb51   not_RB51             456                       456

Documentation

Everything else is in the wiki, whose sources are maintained in docs/wiki:

• Installation — pip, pipx, from a clone, and what gets installed
• Usage — inputs, every option, and the three reports
• Methods — how sizes become canonical bands, how a profile is matched, how the rough flag is read
• The assays — what each resolves, what it cannot, the RB51 junction assay in detail, and the publications
• Validation — every primer checked against its publication and every band size against 87 real genomes, including the primer sets that were checked and rejected
• Large scale validation — all 2051 GenBank assemblies under taxid 234. Of the genomes that produced a call the assay matches the organism in 97.4%, including 19 where it contradicted the submitted label and was right to; zero calls on the 333 Ochrobactrum-derived genomes
• Development — the test suite and the legacy 2023 scripts

All primers and expected band sizes live in brucetyper/assays.py and nowhere else. Edit that file to change how genotypes are called.

Related

• insilicoPCR — in silico PCR for assemblies and reads, with qPCR/probe assays, degenerate primers and a virtual gel. Use it for probe based assays, which brucetyper cannot represent: it reads amplicon sizes the way a gel is read, and a hydrolysis probe has no size to read.

Citation

If brucetyper is useful in your work, please cite it together with the assay publications listed on the wiki:

Duceppe, M.-O. brucetyper: in silico Bruce-ladder, Suis-ladder, AMOS and RB51 junction PCR for Brucella. https://doi.org/10.5281/zenodo.23815138

The DOI above always resolves to the latest release; see CITATION.cff.

License

MIT. Copyright held by His Majesty the King in Right of Canada, as represented by the Canadian Food Inspection Agency.