tb-amr-benchmarking

Pipeline for benchmarking TB AMR callers.

Installation

Nextflow and singularity need to be installed.

Clone this repository, and build the singularity container:

cd singularity
sudo singularity build container.img singularity.def

Those commands made a new container called container.img.

Running the pipeline

The pipeline is run using nextflow. Basic instructions follow, but you may need to change nextflow options or config depending on your environment.

The nextflow script to run is nextflow/run_callers.nf in this repository.

Two input tab-delimited files are required:

• A file with sample names and reads file paths. This should have no header line, and one line per sample. Column 1 = sample name, Column 2 = reads file 1, Column 3 = reads file 2.
• A file defining which callers to use. This needs the following columns:
  1. Name of caller. Must be one of: ARIBA, KvarQ, MTBseq, Mykrobe, TB-Profiler.
  2. 0 or 1. 0 means for each sample, do not rerun if the results are already found. 1 means always rerun, deleting the old data. See the section below on rerunning the pipeline.
  3. Name of tool to use in output files (will also be used as the subdirectory name in the output directory).
  4. Species. Must be "tb" (this is here for the future, when we may want to evaluate other species).
  5. Only applicable to ARIBA and Mykrobe. If ARIBA, should be the reference directory. If Mykrobe, should be a panel name to use a built-in panel, or a "." if providing your own panel using columns 6 and 7. Use "." for all other tools.
  6. Probes FASTA file. Only applicable for mykrobe, otherwise use ".". Must not be "." if tool is Mykrobe and column 5 is ".".
  7. Mykrobe resistance to drug JSON file. Put "." if the tool is not mykrobe. Must not be "." if tool is Mykrobe and column 5 is ".".
  8. Any command line options to pass to the tool when it is run. Put a "." to use the default options.

Assuming those files are called data.tsv and callers.tsv, run the pipeline with a command like this (where you will need to fix /your/path/to/the/container.img):

nextflow run nextflow/run_callers.nf \
  -with-singularity /your/path/to/the/container.img \
  --species tb \
  --input_data_file data.tsv \
  --callers_file callers.tsv \
  --output_dir OUT

That will write the output to the directory OUT.

All the results will be summarised into a single JSON file called OUT/summary.json. This file is used as input to code in the repository https://github.com/iqbal-lab-org/tb-amr-benchmarking-paper.

Rerunning the pipeline

The pipeline can be run more than once on the same output directory, but with different callers (by using a different callers.tsv file), or by adding more samples (by adding lines to the data.tsv file).

• If you are adding new samples, then rerun the nextflow run command with no changes.
• If you are not adding new samples, then add the option --no_new_data to the nextflow command.
• If you have changed the callers.tsv file, then rerun with the option --no_new_data. Or if you have also added new samples, then do not use the --no_new_data option.

If new samples or tools are added, then those new samples/tools will be run. Existing runs will either be left alone (ie not rerun), or will be deleted and rerun depending on column 2 of the callers.tsv file. 0 will mean that nothing is rerun (but new samples/tools are run). 1 will mean that any existing results for that tool are discarded, and everything is run from scratch for that tool.