Skip to content

Starting a run

Now that the configuration and sample sheets are ready, we can start an analysis!

Basic usage

Starting and controlling the workflow execution is done through the snakemake command.

conda activate snakemake
snakemake --use-conda --conda-prefix ~/conda-envs --cores 1 \
  --configfile ~/FooDMe/config/myconfig.yaml

Info

Depending on your system you may want to assign more cores to the analysis. Each additional core will allow running another job in parallel, considerably speeding up execution time. You can adjust the number of assiged cores with --cores N whereby N is the number of cores.

Info

Specifying a prefix for the conda environment is not nescessary but will allow you to reuse created environments between runs. Doing so will save up, time, ressources, and memory on each execution.

Run-specific parameters

As creating or modifying the configuration file to change the sample sheet and the output directory for each run can be a bit cumbersome, it is possible to dynamically modify parameters at execution time. 

WORKDIR=$(date +%F)_foodme
SAMPLES=~/raw_data/samples.tsv

snakemake --use-conda --conda-prefix ~/conda-envs --cores 1 \
  --configfile ~/FooDMe/config/myconfig.yaml \
  --config workdir=${WORKDIR} samples=${SAMPLES}

The above example will use the sample sheet located under ~/raw_data/samples.tsv and outputs the analysis results in the [YYYY-MM-DD]_foodme folder, regardless of the value for the workdir and samples arguments in the config file. The rest of the parameters will be taken form the configuration as usual.

Advanced Snakemake options

A large number of options are available when running snakemake that be can seen using the --help argument or in the online documentation.

A few notables commands are:

  • --forceall will force the recalculation of all the workflow steps. Usefull to repeat an analysis when something went wrong.
  • --notemp will disable temporary file removal. This will considerably increase the size of the output but can be useful to troubleshoot an analysis or puzzling results.
  • --report will output run statistics.