The default layout for the Snakemake working directory, i.e.
the directory specified with the option --directory
/ -d
when invoking snakemake
, is standardized to ease navigating
through the subfolders. In the following, the abbreviation
WD is used to refer to that directory.
If you followed the CUBI recommendation and created your
Snakemake WD with the init.py
script from the workflow
repository, the WD will have been created one level above
the workflow repository location (where the init.py
script resides).
Inside the WD, you find several subfolders ...
The only subfolder relevant to you is called
results/
. This folder contains all important
result files of the analysis, plus two
special files (manifest*
and run_config*
)
that you should never delete, but can ignore.
The entire content of the results/
folder should
be saved in a secure location (with backup). Deleting
any other folder inside the WD does not affect your results,
but may cause delays if the analysis run is changed
or continued later on - so please be sure that the
analysis is complete before deleting anything.
By convention, a pipeline must sort all data files into one of the following folders:
proc/
: processing; intermediate/temp data files, not vitalresults/
: everything relevant to the client/user- automatically contains a copy of the sample sheet, the manifest file, and a dump of the run config
global_ref/
: global reference files either loaded from reference containers or ex nihilo, i.e. references manually placed in the pipeline contextlocal_ref/
: reference data local to the pipeline context, e.g. postprocessed files generated as part of the runlog/
: target for Snakemakelog:
filesrsrc/
: target for Snakemakebenchmark:
files
There are a few other standard directories defined that you must
use if needed. See the Snakemake module workflow/rules/common/10_constants.smk
for a complete list.