Back in 2014 I was working on a large (for the time) set of Target Enrichment panels. For the set up of the panels we wanted to check that the regions we wanted were effectively captured, and I started using some standard tools (like samtools, bedtools…) but that resulted in quite a slow step to be performed several times.

So I asked my talented colleague Giovanni if he could help with his C++ skills, and indeed a program called coverage was born in a matter of a day. Using a library to parse BAM files (libbamtools-dev) I could feed a sorted BAM file, the target BED and obtain a BED output with the covered regions.

Every now and then I asked some new features, namely a lot of statistics to save time doing the maths later (like: multiple BAM file support, % of the target covered 0X, <10X, <20X …). And the program became more and more complex, at the point that during a series of accidents (both of us moving, end of Target Enrichment projects and so and forth…), the source became complex to maintain, forgotten and finally lost.

That became a good lesson for both of us.

From the ashes of that doomed project we summoned covtobed, that we designed:

• To adhere to the UNIX programming style, working with streams
• To be finally easy to install (distributed with Miniconda and Docker)
• Simplified to avoid aving a big blob again

If you want to try it, installing is thus as easy as:

conda install -y bioconda covtobed

And the documentation is now in a GitHub Wiki.