One Day, Many Tool Updates: microGalaxy Hackathon Recap
Eight contributors, one day, and a pile of outdated metagenomics tools — here is what the October 14th microGalaxy hackathon achieved!
On this page
On October 14th, we ran the microGalaxy hackathon we announced a few weeks ago — and it worked out even better than hoped! Eight of us from the microGalaxy community blocked a day, picked a tool from our outdated-tools list, and got to work. The format is simple but effective: no talks, no slides — just everyone updating the tools that our users rely on every day.
Already merged
Right around the hackathon, these updates landed in the IUC repository:
- MEGAN (6.21.7 → 6.25.10), updated and reviewed by Saim
- Tasmanian Mismatch (→ 2.1.0), also by Saim
- StrainDesign (→ 3.2.3), updated by Povilas
- dRep (3.7.1 → 4.0.2) and mmseqs2 (→ 18), merged just in time
Still in flight — reviews welcome!
Several updates are finished or nearly finished and waiting for a review. If you want to help, these are easy wins:
- Bakta (1.9.4 → 1.12.1)
- PICRUSt2 (→ 2.6.3) and DISCO (→ 1.3), both by Saim
- prokka — Povilas even pushed a fix upstream to prokka itself and prepared the wrapper test update for it
- savont, a new tool by Matthias
- GTDB-Tk and argNorm, taken on by Hugo
- PhyloPhlAn, carefully reviewed by Mina
Beyond plain version bumps, Paul started updating antiSMASH to v8.
Automating the boring parts
Not everything in tool maintenance is exciting, which is why Matthias spent part of the day testing gxy-tool-bot on Björn’s galaxytools repository. The bot is an agentic assistant: you file a tool request or an update request as a GitHub issue, it researches the tool and posts a plan, and once a maintainer approves it, it writes the wrapper changes and opens a pull request. In Björn’s repo it worked through a whole batch of wrapper updates — EDTA, fastq_info and CleanLab among them. It will not replace a reviewer, but it takes a lot of the repetitive work off our hands.
Some numbers
We also dug into who actually still uses the older amplicon tools: over the last five years, mothur was run ~870,000 times and DADA2 ~550,000 times on the main servers. So the answer to “is mothur still needed?” is a clear yes — old tools stay relevant as long as published workflows depend on them, which is exactly why keeping them maintained (and reproducible) matters.
Thank you!
Thanks to everyone who joined — Saim, Povilas, Matthias, Berenice, Hugo, Mina, Rand, and everyone who jumped in for reviews and discussions. The full list of tools still needing updates lives in IUC issue #8301, and we coordinate on the microGalaxy Matrix channel. Tool wrapping experience is helpful but not required — there are tasks for all levels, so join us next time!