Entering edit mode
11 days ago
Arthur
•
0
Hi all,
Even though I fully believe that the STAR aligner software is tested against test datasets, and possibly synthetic datasets for accuracy across builds and releases, I cannot find any documentation confirming this?
Is anybody aware, or can anybody point me to documentation that confirms that testing for accuracy and correctness of every build of STAR is part of their CI/CD pipelines?
Many thanks in advance!
Just curious: why do you want this documentation and why are you singling out STAR? Can you find this documentation for other bioinformatics software (cutadapt, hisat, bowtie, bwa, minimap, trinity) or is STAR the only bioinformatics software you’ve used that lacks this documentation?
Fair question - I single out STAR as I believe it's a widely used tool, but I am curious across the board of these 'widely used bioinformatics tools' how the quality is managed. Coming from a software engineering background, I believe that all these tools must have some sort of quality assurance as part of their build process, but it seems that not everybody agrees with that assumption. That's why I'm curious to see if there is indeed any evidence that 'widely used' bioinformatics tools have such mechanisms in place inside their build processes.
Curious as to how many people were part of your team in (I assume) commercial software dev?
Most open-source bioinformatics software is developed by small teams of contributors (or in some cases single individuals). All of them must have an internal set of regression tests that they run a build through before releasing new versions. Since documentation is likely the last thing that a open source developer wants to tackle, this information remains lacking for many packages.
We have a number of developers who participate here regularly and perhaps one of them will chime in and comment about the process they internally use to ensure consistency.
Apologies in advance since I will probably miss some devs :
Alex Reynolds ; chrchang523 ; Brian Bushnell ; shenwei356; Pierre Lindenbaum