Documenter légèrement BAM
However I think this should remain a hidden feature for two reasons:
A BAM file is generally obtained by aligning reads to reference sequences. This could make people think they have to align their reads beforehand.
In the case of paired-end sequencing, there is only one BAM file (but two FASTQ files). So the people could provide a single BAM file where reads have not been merged.
On pourrait mettre exactement ce type d'infos dans
doc/algo.org, et ne pas en faire la pub dans
doc/user.org et sur le serveur.