The goal is to have distinct counts for -5p and -3p miRNAs.

This caused bugs with libraries named "1_1", which was interpreted as
number 11.

Could happen with columns containing only zero counts.

This makes it easier to update dependencies (libbamutils is a dependency
of libhts).

Version numbers should be strings.

Don't forget setup.py.

There was a failure to install pyseqlogo from github.

See https://stackoverflow.com/q/71504533/1878788

Maybe it would be good to try other packages to plot logos (biotite ?).

This should be done via a "nodedup" switch in the pipeline config.

This resulted in a wrong symbolic link in the container.

Also made the genome preparation repository a submodule.

It has been generated using https://codemeta.github.io/codemeta-generator/, and is meant to help submission to Software Heritage.

This is part of an attempt to submit the code to Software Heritage
(https://www.softwareheritage.org/save-and-reference-research-software/)
to obtain a DOI for Eric Cornes' paper.

Function writing fastq was likely prematurely closing files due to wrong
sRNA name (tRF instead of tRFRNA).

Not tested yet.