Sailing in @biorxivpreprint.bsky.social ready for summer, our study on the iconic Portuguese man o' war 🪼 is out: www.biorxiv.org/content/10.6.... We use (epi)genomics to understand a critter that fascinates me since childhood. Massive team effort with @obog.bsky.social, Cummins and Neely groups. 🧵
James Ferguson
@psy-fer.bsky.social
Bioinformatician/Genomics Software Engineer @garvaninstitute.bsky.social Views my own. Mastodon @Psy_Fer_@genomic.social, https://genomic.social
zooid specialisation is up there as one of the coolest things i've ever encountered in biology. Read all about the genomics involved in specialisation and the molecular basis of venom activity in the Bluebottle Physalia Utriculus
Integrative genomics of the siphonophore Physalia utriculus reveals the regulatory logic of colonial division of labour and the molecular basis of venom activity https://www.biorxiv.org/content/10.64898/2026.08.01.742136v1
A scientific plotting library in Rust. 60 plot types, SVG output, optional PNG/PDF backends, and a CLI binary that renders plots directly from the shell — including in the terminal itself. github.com/Psy-Fer/kuva
I was recently invited to provide a case study along with other researchers for this blog post+pre-print. My main points are: AI agents are useful, but steering through expertise, verification, responsibility, and support are more important than ever openai.com/index/scient...
Scientific computing in the age of agentic AI
A new field report shows how scientists use AI coding agents to modernize scientific computing, accelerating software development and discovery in genomics and beyond.
openai.com
It was fun being invited to contribute a case study to this #OpenAI preprint + blog 😊 I think it ended up a really nice overview, an interesting cross-section of how people are approaching similar problems with #ai #agent -assisted #bioinformatics software projects...💡 openai.com/index/scient...
Scientific computing in the age of agentic AI
A new field report shows how scientists use AI coding agents to modernize scientific computing, accelerating software development and discovery in genomics and beyond.
openai.com
Coding agents are helping scientists spend more time advancing research, taking on everything from routine maintenance and targeted optimization to complete redesigns and new systems. (1/2)
AI in biology must be experimentally verifiable, and that requires wet labs. @astar_gis is hiring pre- and postdocs with proven AI knowledge to build foundation models for RNA therapeutics: mRNA vaccines/drugs, gene editing, ribozymes and in vivo CAR-T. Singapore-based. DM me and please repost.
Cool new feature for kuva i've been cooking up. So when you use the CLI, you can get a nice plot, but what if you want to now port that over to the library? well, --emit-code will dump the rust code needed to re-produce the same CLI built plot. This will be in the next release :)
kuva v0.4.0 released Black & white / accessibility mode across all 61 plot types, sharper text layout, and axis dead-space fixes. A big thanks to @tfenne.bsky.social for the great bug reports and PRs this release. github.com/Psy-Fer/kuva
Fable 5 might be good, but I wouldn't know, because I can't use it in my personal or work accounts, because no matter what I try to do with it, it trips the safeguards and downgrades to opus 4.8
Someone I know vibe coded an app. Basically a todo app with lots of integrations and they spent a few days adding data from their life. Now vibe coded app timelines are short. So short in fact, they didn't realise their database was at /tmp/data.db till it was too late. oops 😂
How good is MiniBWA, the successor to BWA? To test it, I ran MiniBWA on sequencing from 76 different species, comparing mapping speed, rate and accuracy with BWA MEM. In short, it's really good. If you map short reads, it's well worth your time. andrewcarroll.github.io/2026/06/30/t...
The Best of Both Worlds - Assessing MiniBWA
Recently, Heng Li released MiniBWA (GitHub) alongside a paper by Heng Li and Nils Homer describing the method (paper). MiniBWA builds on the approaches in Minimap2 (also by Heng Li), but falls back on...
andrewcarroll.github.io
Sequencing is not just a test. It is the front door to the next cancer system. My OpEd in The Australian is online today, ahead of tomorrow’s UNSW Showcase at Parliament House. 🧵⬇️ www.theaustralian.com.au/health/medic...
kuva v0.3.0 release! - New raster backend: faster than the old one, dropped dep. - New QuiverPlot type - Pre-compiled release binaries (thanks @wdecoster.bsky.social ) - .parquet input for all CLI subcommands Also lots of fixes and other features crates.io/crates/kuva github.com/Psy-Fer/kuva
going through all the last checks for a new kuva release, and gosh some of the test plots are pretty 😍
salmon 2.0 (the rust re-write) is now live! It's also the version you get if you install via bioconda. This is a from-scratch rewrite in rust, and the base for future development. I'll post a more detailed thread tomorrow, but you can read the docs here combine-lab.github.io/salmon/
salmon
Fast, accurate transcript quantification from RNA-seq reads.
combine-lab.github.io
lookout, i'm asking the tough questions, to fable, and it's not having it!
Glad to announce that our new long-read metagenomic SNP caller, SNooPy, is published in NAR! Existing long-read SNP callers are not designed for metagenomic data, check out our new solution 👉https://academic.oup.com/nar/article/54/10/gkag556/8700491 @narjournal.bsky.social
Jeremy Wang developed rammap, a minimap2 rewrite in Rust. It achieves comparable or better performance than minimap2 and produces identical output to minimap2. During rewrite, Jeremy found two long-existing bugs in minimap2 which are fixed in v2.31. www.biorxiv.org/content/10.6...
biorxiv.org
In case you don't follow it already, reminder that this exists! scientificcomputing.rs/monthly/
Scientific Computing in Rust
the Scientific Computing in Rust annual workshop and monthly newsletter.
scientificcomputing.rs
svb v0.2.0 released! Added SVB-ZD, the slow5 format signal compression prep method. (before it goes into zlib or zstd) Again, I think you can see where this is going 😅 github.com/Psy-Fer/svb
GitHub - Psy-Fer/svb: streamVbyte in pure rust for u16/u32/u64 plus extras
streamVbyte in pure rust for u16/u32/u64 plus extras - Psy-Fer/svb
github.com
Memorial Day weekend project: port CaPS-SA parallel suffix array construction algo to Rust (crates.io/crates/caps-sa)! I also have a branch to integrate this into the ruSTAR-aligner, getting a *byte-for-byte* STAR compatible index on the human genome with just 10.7GB of RAM in ~18min (32 threads).
crates.io: Rust Package Registry
crates.io serves as a central registry for sharing crates, which are packages or libraries written in Rust that you can use to enhance your projects
crates.io
Testing SIMD across different architectures is tricky. Anyone got a solid way of doing it without just having multiple PCs with all the various CPUs in them available? maybe github runners? (though not good for benchmarking)
A love letter to @nanoporetech.com pod5lib for reading and writing wire compatible pod5 files in Rust. Compared to the python/C++ library with 50k reads it is: - Read: 1.9 faster and uses ~60x less RAM - Write: 1.7x faster and uses ~5x less RAM github.com/Psy-Fer/pod5... crates.io/crates/pod5lib
GitHub - Psy-Fer/pod5lib-rs: pod5lib in rust
pod5lib in rust. Contribute to Psy-Fer/pod5lib-rs development by creating an account on GitHub.
github.com
So I was thinking about streamvbyte encoding used in ONT signal compression, and the delta/zigzag steps too, before it goes into zstd compression. And I noticed that while there were a few crates in rust for u32 and u64, there wasn't one for u16 (what is used in VBZ for ONT data) So I made one
@nilshomer.com joined Rob Patro, @ewels.bsky.social, & others to talk AI-assisted bioinformatics rewrites on the OMGenomics show. The AI-assisted rewrite is the easy part. It is proving correctness, finding edge cases, and being around to support the tool later that’s harder. youtu.be/0o2XnEBDxrI
I interviewed 5 bioinformatics experts about AI rewrites
YouTube video by OMGenomics
youtu.be
Iterating through various raster methods for drawing lines for making PNG files generate faster with kuva on very large dataset inputs. Made some good progress. While the polyline-opt method was faster, it's visual fidelity was...ahh...shit. So doesn't count haha
me: Okay, time to update the png/raster backend of kuva to be faster ... me: aaand done...wait...what? HAHAHAHAHAHAHAHA
In the kuva v0.2.0 update, while I was building the venn diagram plot type, I wanted a way to visually see the overlap group membership that wasn't as ambiguous. So I added these coloured membership dots to help with that. psy-fer.github.io/kuva/plots/v...
kuva v0.2.0 Now over 60 plots! New plot types: Gantt, Network, Waffle, Horizon, PopulationPyramid, Joint, Clustermap, ROC, PR, Lollipop, Survival, Raincloud, Slope, Venn, Parallel, Mosaic, Streamgraph, Radar, Treemap, Sunburst, Bump, Funnel, Rose, Calendar, Legend, Text github.com/Psy-Fer/kuva
I am convinced that certain forms of techno/hardstyle/etc are inspired by MRI machine noises.