Should we cite sed / awk in papers?

Question

[amy__]

Hi all,

Not necessarily bioinformatics based but for some of my bioinformatic pipeline intermediate steps I've used awk or sed to perform file manipulation. In a paper or a thesis, would you cite awk and sed?

I don't think I can recall seeing it before in a paper, but I thought I'd ask for your opinions!

Thanks, Amy

[jkbonfield]

Generally I've never bothered citing computer languages or fundamental tools like that. If it comes preinstalled with the OS, that's a pretty good indication it's not needing citations any more, although it's still worth documenting somewhere.

However if it was a non mainstream language or being supported by a single group whose work depends on grants then yes I would definitely cite them. Anyone who works in a scientific field whose career may depend on rather blunt metrics such as paper citation counts deserves to be cited, even if they're getting many thousands of them! (I've been at both ends of the spectrum for this, with minimally cited work and very well cited work. It definitely did matter.)

I've also sometimes cited internet RFCs where a file format specifically utilises a compression protocol or a data-encoding format. That's part of the methods section IMO.

[GenoMax]

If you used a variant of awk (mawk,gawk etc) you would want to make a note of that.

Answer 1

[ATpoint]

No, these are lowlevel Linux applications that do not need citations. If you do then also your email client and the coffee machine needs citations (pun intended). Thinking about it, I feel the coffee machine had a larger influence on my PhD thesis then sed, so maybe indeed it should have been cited. But seriously, in a thesis you can do whatever you want, but in a paper often you have citation and word limits and wasting them on sed and awk would be the last thing to do.

[amy__]

Perfect, thank you lots!

[ATpoint]

That having said, documenting exact software versions, at best via some sort of container, less accurate being a conda environment, even less a list of software versions, worst non version control at all, and code tracking via Git(hub) should be mandatory and key for reproducibility, and be it for your own sake alone.