#kerblam — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #kerblam, aggregated by home.social.
-
I've updated Kerblam! to v1.2.0!
Some new features, some new bugfixes, all in time for the holidays!
-
To fix an issue I did a huge refactoring of #kerblam and now the code is so much nicer.
Shoutout to #rust for simply *working* after all the imports were fixed. The moment the compiler was happy all tests just magically passed with no problems. I moved basically everything around, and the fact that nothing broke is incredible.
-
The preprint for #kerblam is online at https://doi.org/10.48550/arXiv.2410.10513
If anyone would like to #peerreview us I'd love any feedback before we send it to some journal or other.
We talk about the structure of #dataanalysis projects, #FAIR, #OpenScience, #reproducibility and all the good stuff.
If you *do* have comments, you can contact me on here, or just anywhere (check my bio).
Thanks!
-
#kerblam hit 10 stars on Github baby! I'm basically famous. Please, hold your applause.
-
I don't want to brag, and this account is basically ads for Kerblam, but I did a differential expression analysis for some colleagues in like an hour today using it. New project, done, fetch data, done, write tiny r script, done, docker container for the dependencies, then just kerblam run, data package and export the pipeline with one command.
I must say, I'm very pleased with myself.
-
Kerblam! version 1.0.0 is finally out!
If you don't already know from my infinite nagging, Kerblam! is a project manager, dealing with organising your data analysis project's filesystem, input/output/intermediate data and workflows. It also helps when sharing your work with others.
Check out https://kerblam.dev/ to get started!
Kerblam is FOSS and open to contributions: https://github.com/MrHedmad/kerblam
#kerblam #DataAnalytics #rustlang #productivity #DoctorWho (Heh)
-
I've stopped making new releases to #kerblam but there are new features on the main branch!
https://github.com/MrHedmad/kerblam
In particular:
- You can now pass extra arguments to make or bash with kerblam run pipeline -- my other args
- A long-running glitch wheras runtime profiles triggered make builds needlessly was fixed (just now!)
- There were a bunch of edits of the documentation and overall spit polish;If you don't know anything about kerblam, you can get an overview on https://kerblam.dev/
-
We have sent #kerblam (https://github.com/MrHedmad/kerblam) to *bioinformatics* for #peerreview.
I have little hope of it being accepted, but it's been under review for a month and a half now, and I cannot release new versions (even though I've added more features and bug fixes) because we are under review.
Journal-based publishing is just incompatible with modern science. We have to come up with something better.
-
I just released #kerblam version 0.3.0 "A light in the dark". Other than a few bug fixes it includes a way to add descriptions of your pipelines so that you don't forget what they do.
Plus, a few quality-of-life changes, like showing you a list of pipes when you do not specify one (just like you did when you misstyped one)
-
#kerblam is viable! I'll make the first beta release tonight.
If you do #datascience, #bioinformatics or any sort of data processing that needs to be reproducible, perhaps kerblam will be useful to you!
Let me know of any problems or thoughts that you can think of! I'd greatly appreciate any feedback.
(github stars also make my day if you want to leave one thanks luv you) -
#kerblam compiled with debug information (e.g. `cargo build`): 127 Mb.
Compiled with `--release`, Link time optimizations and stripping symbols: 4.5 Mb.That's more than a x38 smaller binary! Neat!
-
`kerblam package` is implemented! It works on my (small, easy) examples, so it should be ready for more live testing.
It packages pipelines and the minimum amount of data your project needs for execution into a container, so you can share a ready-to-run container for others to reproduce your analysis.
It should make it easier to share computational pipelines with others, for e.g. preprints or with colleagues. -
#kerblam now still cleans up after itself even if you kill your pipeline with `CTRL-C`. It will abort entirely if you press `CTRL-C` twice in a row, if the child process is stuck, allowing the OS to deal with the orphan.
-
So, turns out that #kerblam run worked only for the simplest of cases but I fixed it now!
On the few manual tests that I've done, shellfiles, makefiles and dockerized runs now work fine, and work nicely with profiles. Rust makes it all super fast too, seemingly with no overhead at all!
With kerblam you can run different makefiles/shell files with your pipelines, wrap them in docker, and run them on different data in a jiffy!
-
`kerblam run` (mostly) works!
With it, you can dynamically run different makefiles with different input data based on profiles, so you can break down different aspects of your #bioinformatics or #datascience analysis and run them with ease.
I'll iron out the bits that don't work tomorrow.
I'll post more "ads" for kerblam as I develop it. Mute #kerblam if you don't want to see them!