GitLab

Use 8 cores instead of just 1, according to the documentation speed
should almost scale linear with the cores provided.

Also reduce the compression level on the output file, since most time
for cutadapt is spent re-compressing the data after trimming the
adapters.

Revert "Remove explicit tmp folder"

See merge request !17

This reverts commit 1b7d807f

This should no longer be needed on the slurm cluster, where each task
can request the amount of tmp space it requires explicitly.

Removing the tmp from the shared filesystem back onto the host running
the analysis should also improve the performance of the pipeline.

This way, the same version of picard is used in all tasks in the
pipeline. This fixes issue #38

Add optional bed coverage output files

See merge request !16

Unfortunately, this adds a limitation on the sample names that can be
used with Hutspot, since the naming of the samples in the multiQC parsed
output of picard MarkDuplicates is partly ambiguous.

This limitation has been added to the readme, and a check has been added
to the pipeline snakefile to throw an error when overlapping sample
names are detected.

Previously, Hutspot supported multiple bed files to calculate coverage
against. Because of this, the stats.json file had a nested structured
where the coverage based on each bed file was stored, including the name
of the bed file and the gender, according to that bed file.

Since the current version of Hutspot only supports a single bed file,
this structure has been simplified. All coverage statistics are now
directly under 'coverage' for each sample, and the 'gender' has been
moved out of the 'coverage' statistics to the sample level.

Hutspot now supports two bed files to calculate coverage. One is the
`targetsfile`, which was called `bedfile` before, and holds the targets
of the capture kit. The other one is `baitsfile`, which holds the bait
locations of the capture kit.

It is possible to specify only the `targetsfile`, but if you specify the
`baitsfile`, the `targetsfile` must be specified as well, since both are
required by picard HsMetrics.

Also added a test for invalid configuration files, and shortened the
jsonschema validation error to only show the human readable message.

Pass 'no file' to collect_stats.py by using an empty list in the
Snakefile and nargs='?' in the python script. This is cleaner than using
'.' as a special file and parsing that logic in the collect stats
script.

If both a target and bait bedfiles have been specified, calculate the
hybrid-selection (HS) statistics using picard.

Multiple (pytest) processes trying to write the same image to
/tmp/singularity can lead to corruption, leading to intermittent
failures in the gitlab-ci tests. By specifying a singularity prefix in
the snakemake profile, the same images can be re-used, so we only have
to worry about concurrent processes writing the same image when a new
image is added to the pipeline.

This also required reordering some snakemake rules to make sure that the
correct input files are available. When using rule based inputs, the
rules in the Snakefile have to be sorted, and only rule based inputs
from rules that occur earlier in the Snakefile can be used.

The implementation is a bit hacky, since snakemake does not allow for
optional input files. As a workaround, "." is passed when the bedfile is
not defined, and the collect_stats.py script has been made aware of the
special meaning of "."

Additionally, Click has been removed as a dependency for collect stats,
and the structure of the stats.json file has been updated to only allow
for a single entry of coverage stats instead of a list. This has been
done to match an earlier change in Hutspot where support for multiple
bed files has been dropped.