Now I’m cashing in our parasocial contract. My New Year’s resolution is to quit social media (except Bluesky, obviously). Wish me luck! Hopefully it will go as well as my 2025 resolution: one full year without alcohol!

Consider signing this petition regarding restoring free access to the ACM digital library: www.ipetitions.com/petition/res...

Liminal format

A three act thread dear to my heart

The correct way should be the left one though

Looks like we might have the first time AI directly solved an open Erdos problem: github.com/teorth/erdos... A couple times recently this appeared to be the case only to turn out to be a successful deep lit review by AI.

Sequencing Instrumrnt Outlook 2026 My observations & predictions on sequencing instrument companies 🧬🖥️ omicsomics.blogspot.com/2026/01/sequ...

How delightful to start the year with a neat overview of the sequencing field from the most capable watcher of the space.

Now published in Algorithms for Molecular Biology: link.springer.com/article/10.1.... Key message: a tiny CNN model with 7k parameters can capture main splice signals across vertebrates+insect and halves the minimap2 & miniprot junction error rate. I always use this new feature now.

How well does TWAS estimate a gene’s direction of effect on a trait? We think of this as an important stress-test for the accuracy of TWAS. In a new pre-print, we find that TWAS gets the sign wrong around 20-30% of the time! doi.org/10.64898/202... 1/n

Le crédit impôt recherche aura couté 7,8 Md€ en 2023, soit presque 2 fois le budget du CNRS (4,1 Md€). De l’argent donné à des multinationales sans aucun vrai contrôle de la recherche scientifique effectuée. www.aefinfo.fr/depeche/7427...

If you're not aware, essentially the deal is that the twitter AI, Grok, has a pretty good image model. So, someone posts a photo, and you can say "Grok, show this person in a bikini" and get realistic results. I don't think actual nudity, but certainly highly sexualized clothing.

A thread 🧵 of _terrible_ design decisions for the internal APIs of the two Italian high speed train operators, Italo and Trenitalia. Hardcoded passwords, terrible translations, spelling mistakes, xml in json (but utf8 encoded and transmitted as an array of numbers), it's all here. Enjoy!

Strobealign can now map long reads. Still POC, i.e., PAF only -- no supplementary chains or piecewise extension (yet). 3/4

Strobealign v0.17.0 is out! github.com/ksahlin/stro...

Some bioinformatics papers: this lightweight method requires 4 gigabytes of RAM. Other bioinformatics papers: this lightweight method requires 4 terabytes of RAM.

A powerful and unbiased way to identify links between genetic variants and biological traits has changed scientists’ understanding of complex diseases and how to treat them go.nature.com/4s1lL3M

XTree enables memory-efficient, accurate short and long sequence alignment to millions of genomes across the tree of life https://www.biorxiv.org/content/10.64898/2025.12.22.696015v1

I was looking for a crate to measure raw perf counters from rust for so long, and then out of the blue @rickbitloo.bsky.social dropped perfcnt on me! Super hyped to start using this! We already made some nice plots where we correlate walltime improvements with higher IPC. docs.rs/perfcnt/late...

Nature research paper: An 11-qubit atom processor in silicon go.nature.com/3Y4fN4m

The paper writing season is here!

Le choix de quitter twitter pour bluesky aura cruellement démontré que la majorité préfère s'accomoder du fascisme plutôt que de devoir changer ses habitudes.

Closing out my year with a journal editor shocker 🧵 Checking new manuscripts today I reviewed a paper attributing 2 papers to me I did not write. A daft thing for an author to do of course. But intrigued I web searched up one of the titles and that's when it got real weird...

I have become a slop collector. 😮‍💨 Dirty job, but someone has to do it. I’m compiling gen A.I. images in academic journals. This is an ongoing project, and version 1 is here: doi.org/10.6084/m9.f... Version 2 in January with at least three 3️⃣ new illustrations. #AcademicChatter

What

“This year, renewables surpassed coal as a source of electricity worldwide, and solar and wind energy grew fast enough to cover the entire increase in global electricity use from January to June according to energy think tank Ember“ @science.org’s Breakthrough of the Year is the growth of renewables

No signs of an end to rapid gains in AI ability at ever-decreasing costs (which is a log scale) yet. I have to update this monthly or more frequently at this point. All AI benchmarks are flawed, but GPQA Diamond has been a pretty good one, though likely close to being maxed out.

De novo genome sequence assembly of the model algal endosymbiont Micractinium conductrix derived from its host Paramecium bursaria 186b. Guy Leonard, Irma Vitonyte, Fiona R Savory, Erika M Hansson, Duncan D Cameron, Michael Brockhurst, Thomas A Richards bioRxiv doi: doi.org/10.64898/202...

Transcriptomics in the era of long-read sequencing #Review by @carolinamonzo.bsky.social, Tianyuan Liu & @anaconesa.bsky.social @conesalab.bsky.social @csic.es Free to read here: rdcu.be/efBjP

GSC: efficient lossless compression of VCF files with fast query academic.oup.com/gigascience/... 🧬🖥️ github.com/luo-xiaolong...

SAMtools / HTSLib / BCFTools 1.23 has been released. See the releasenotes on GitHub and www.htslib.org for details.

And it's posted! If you're interested and eligible, please consider applying through the UMD portal: umd.wd1.myworkdayjobs.com/en-US/UMCP/j.... If you're a PI working in algorithmic genomics (& you can recommend my lab to your top graduating students ;P), please let them know!

"whether LLMs can make “scientific discoveries” is no more interesting a question than whether Galileo’s telescope really saw"

💾 Prokka 1.15.6 is released! This is the last major release of Prokka. But don't be sad, because @oschwengers.bsky.social already has an excellent replacement called Bakta you can migrate to. #bioinformatics #microbiology #genomics github.com/tseemann/pro...

Pyranges v1: a Python framework for ultrafast sequence interval operations www.biorxiv.org/content/10.6... 🧬🖥️🧪 Docs: pyranges1.readthedocs.io/en/latest/ Rust backend: github.com/pyranges/rur... 🦀

We're organising a microbes & deep learning session at SMBE next year -- looking forward to seeing your abstracts!

Preprint Alert! With @tmthrz.bsky.social and @rayanchikhi.bsky.social we aim to tackle practical unitigs compression! A thread:

🇸🇰 Slovakia has become the newest member to join #ELIXIR, following the signature of the ELIXIR Consortium Agreement by the Ministry of Education, Research and Youth. 🙌 This brings the number of ELIXIR members to 23. Read the news 👉 https://loom.ly/h4CyhJo

Have you ever wondered how new DNA methylation patterns are established? Paradigm shift ahead! We discovered a new mode of DNA methylation targeting in plants that relies on transcription factors and sequence motifs rather than chromatin modifications to regulate the methylome. rdcu.be/eQ6L5 1/8

I really believe that if people knew how cheap it is to verify thousands of fake accounts for a platform like X or TikTok and properly absorbed that knowledge, they would treat the online information environment very differently...

Leaders at @ukri.org should be seriously looking at the lottery first idea. 68% lower economic costs, increase in diversity of award holders and well received. www.nature.com/articles/s41...

Whoa. This new GDPval score for GPT-5.2 is something. GDPval is probably the most economically relevant measure of AI ability, suggesting that in head-to-head competition with human experts on tasks that require 4-8 hours for a human to do, GPT-5.2 wins 71% of the time as judged by other humans.

My thoughts on the matter: www.nature.com/articles/s41...

Rust 1.92.0 has been released! 🌈 🦀 This version adds RwLockWriteGuard::downgrade(), adds Location::file_as_c_str(), improves diagnostic and validation for various attributes, no longer warns for unused Result<(), Infallible>, and more! Check out the announcement: blog.rust-lang.org/2025/12/11/R...

A database of 2.75 billion buildings could help scientists to monitor urban planning, climate change, disaster risks and even corruption go.nature.com/4acdXWm

The GTDB website now has an ANI calculator based on skani that supports uploading of user genomes. Try it at gtdb.ecogenomic.org/tools/skani. Find more information about @jimshaw.bsky.social fantastic tool at www.nature.com/articles/s41....

My first Rust toy tool (just for learning purposes)!!! The performance (both time and memory) is good! Rust is hard to learn, and I still need more practice! github.com/shenwei356/f...

I’m recruiting a postdoc to work on algorithms for cancer genome reconstruction. We have access to a rich set of tumour samples sequenced across multiple technologies. If interested, feel free to DM. Please share.