Home

Mykrobe

http://www.mykrobe.com

The following help pages are available:

• Installation instructions
• Information on built-in variant panels for mykrobe predict
• How to run AMR prediction (the most common use of mykrobe)
• An explanation of the output from running AMR prediction
• Some notes on Mycobacteria taxonomy and names
• How to cite mykrobe (both the tool itself and publications for the panels)

Advanced topics:

• How to use custom variant panels
• How to do custom lineage calling
• How to genotype a list of variants and/or presence of sequences (no AMR or lineage calling)
• If you made your own panel and have a problem with missed calls: use background variants in your panel.

Quick start

Install

Install mykrobe:

• bioconda - conda install -c bioconda mykrobe
• from source - pip3 install .
• or using singularity or docker (see wiki for details)

Since microbe code versions and panel data are independent, make sure the panels are installed/up to date:

mykrobe panels update_metadata
mykrobe panels update_species all

By default, the panel data is installed in the same place as mykrobe. If you do not have write access, then add the option --panels_dir Panels, changing Panels to the name of a directory that will be created and used to store all panel data. If you do this, then you will also need this option when running mykrobe predict.

Run AMR prediction

Run on Mtb, making a JSON file of results:

mykrobe predict --sample my_sample_name \
  --species tb \
  --output out.json \
  --format json \
  --seq reads.fq.gz

For other species, change tb in the option --species tb to one of: paratyphiB, sonnei, staph, typhi.

Test on example data

You can test that mykrobe predict runs as expected by downloading and running on a small toy set of reads, as follows.

wget -O test_reads.fq.gz https://ndownloader.figshare.com/files/21059229
mykrobe predict -s SAMPLE -S tb -o out.json --format json -i test_reads.fq.gz

The test reads are simulated, and perfectly match the reference, except for the isoniazid resistant associated variant inhA I21T. You should see a section like this in the output file out.json:

"Isoniazid": {
    "predict": "R",
    "called_by": {
        "inhA_I21T-ATC1674262ACT": {
    ... etc