Dunno if this is the right channel, but anyway. I wrote kuva ( https://github.com/Psy-Fer/kuva ) It's a scientific plotting library in rust, but also has a CLI binary mode where you can pipe in data and get plots. It can also plot to the terminal (handy for quick sanity checks on a remote system). It's got minimal dependencies and is about to have a pretty large update (v0.1.7) that will bring the plot type count to ~60. There are some bioinformatics specific plot types in there, like manhattan, volcano, and dot plots, as well as all the regular kinds of plots you would expect. It might be a handy tool for creating plots as QC steps in nextflow pipelines, because it's a CLI tool, which is also super easy to install. I'm always looking for feedback on features, bugs, ergonomics, etc. I try to make the default plots as easy to build as possible so they don't need a lot of boilerplate to get started (in rust, even easier with the CLI) Anyway, let me know here, or in the issues on the repo Here is a collection of plots with all the new ones coming in the next version too. (svg so you can zoom, and a crappy compressed version if you don't wanna downlaod that) oh, it can also play doom in an SVG (https://psy-fer.github.io/kuva/doom.html)

Wow, cool! I've been missing Plotly in rust and struggling a bit, this looks super nice 🤩

Kuva does svg as its core output. PNG and pdf are features you can turn on.

Yeah MultiQC kind of needs all the whizzy interactive HTML stuff which complicates matters.

I have some wizzbang built in interactive components in kuva that I have been working on

Sounds similar to our

nextflow run doom

😅

Yea SVGs are so good. Should be the default image everywhere. No idea why people prefer png/jpeg when svg is vector, can be smaller, and has way more cool stuff you can take advantage of in the spec. I do have plans to make kuva compatible with wasm and expose the render with a public API so people can work with it directly and skip the format specific backends. That let's you embed it in GUI/web stuff.

Yes though for plots, SVGs can also be huge if you have bazillions of points 😅 Can do deduplication based on coordinates to solve that though (just not in MultiQC again, as someone might filter away all points except that one sample right at the back)

This is why things like joint plots and rainclouds/strip/beehive plots are handy. Because you can do dedup and still show the full distributions 🙂

@Phil Ewels did you ever play with https://hvplot.holoviz.org/en/docs/latest/index.html

Nice! Those kinds of libraries are great if you're doing a lot of different plots (eg. a notebook / exploratory analysis etc) 👍🏻