Widespread data leakage inflates performance estimates in cancer drug response prediction https://www.biorxiv.org/content/10.64898/2026.02.05.704016v1

Science should be machine-readable https://www.biorxiv.org/content/10.64898/2026.01.30.702911v1

This sort of take — although generally common — is especially entertaining to read two weeks after FDA guidance on Bayesian methods in clinical trials

Version 3 is out, read it at www.biorxiv.org/content/10.1...! New additions: • Human genomics • Pig transcriptomics • Bottom-up reanalysis of long-read human data

Heavy on potatoes and ammonia

The January issue is live nature.com/nbt/volumes/44 /issues/1 On the cover, Luebbert et al. present a method to detect viral sequences in bulk and single-cell transcriptomic data using conserved amino acid domains instead of annotated reference genomes go.nature.com/4lGrSY3

Systematic cell-type resolved transcriptomes of 8 tissues in 8 lab and wild-derived mouse strains capture global and local expression variation www.cell.com/cell-genomic... @ucirvine.bsky.social

Repeat after me: GENE 👏 EXPRESSION 👏 PROFILING 👏 DOES 👏 NOT 👏 A 👏 VIRTUAL 👏 CELL 👏 MAKE 👏

No contest. Just read the first two sentences of the abstract. www.nature.com/articles/s41...

New #ebook at Project Gutenberg: Pan's garden by Algernon Blackwood https://www.gutenberg.org/ebooks/77472

The mechanics and physics of tofu: Understanding hydrated soft solids through feature networks https://www.biorxiv.org/content/10.64898/2025.12.10.693552v1

lol #qveenherby

My student Catherine Felce will be defending her thesis @caltech.edu next week. If you're in the area consider attending; it will be a treat! Cat's recent work: journals.aps.org/pre/abstract... www.biorxiv.org/content/10.1... www.biorxiv.org/content/10.1...

A Comparison of Computational Methods for Modeling Stochastic Collaborative DNA Methylation Dynamics https://www.biorxiv.org/content/10.1101/2025.11.25.690568v1

Wth @nature.com? www.nature.com/articles/s41...

We've got a really cool preprint out in collaboration with @lpachter.bsky.social's awesome student Cat Felce. Using biophysical models and RNA-seq data, we explore the mechanisms of selective constraint on mRNA abundance, finding constraint on decay rate www.biorxiv.org/content/10.1...

"Move fast with stable infrastructure" is when both your website and your messaging app are so unreliable as to be practically unusable

With its various and playful explorations of authenticity — "found documents", metafictional games with readers, etc. — The Private Memoirs and Confessions of a Justified Sinner (1824) by James Hogg, anticipates some of postmodernism's best trickery: http://buff.ly/2p3UZIt

The challenges of single cell transcriptomics on difficult human tissue: the placenta #SingleCell 🧪🧬🖥️ https://www.biorxiv.org/content/10.1101/2025.11.14.688483v1

Joint Biophysical Modeling of Paired Single-Cell RNA and Protein Measurements https://www.biorxiv.org/content/10.1101/2025.11.14.688548v1

Monod fits biophysically motivated models to single-cell transcriptomics data, providing insights into gene expression dynamics. @goringennady.bsky.social @lpachter.bsky.social www.nature.com/articles/s41...

Tech Innovation from Google (An Alphabet Company), Powered By AI one assumes

This is one of the most remarkable academic debacles I've ever seen. A large RCT got published in BMJ. There are currently 44 Pubpeer comments, mostly about the data, including...well. Read for yourself. pubpeer.com/publications...

We publish *imaginary* research abstracts! The Journal of Imaginary Research is now accepting submissions of new short fictional pieces. Click to read: Why Abstract fiction? The theme How to get started What to submit to us What *not* to submit journalofimaginaryresearch.home.blog/2025/10/28/c...

the unfortunate truth re: data visualisation is that you really do gotta hand it to bar charts

I know I posted this on Twitter ages ago, but I came across it again last night and thought some may not have seen it yet. (From the Not the Nine O'Clock News spin-off book, 1980.)

I have found the candles

Programming skill is just a set of rules you've inferred via personal suffering

To achieve good velocity embeddings, test on training data

iCloud has a fun Easter egg where it deletes all your books from the local device at random intervals so you can't use Apple Books when you get on a plane

Hungarian writer László Krasznahorkai, known for philosophical and bleakly funny novels, has won the Nobel Prize in literature.

"We're going to create superintelligence" How about making outlook search work first. How about that

A new revision is out! Lots of interesting cross-species data analysis, all of it pointing to the same conclusion: male XIST expression in the heart really is only detectable in humans!

scFLUENT-seq reveals that individual cells—lymphocytes to stem cells—transcribe only ∼0.02%–3.1% of the genome, versus >80% in bulk, revealing limited genome engagement and profound cell-type and cell-to-cell heterogeneity @cellcellpress.bsky.social www.cell.com/cell/fulltex...

the lengths Apple will go to to avoid making consumer software are amazing. hope we get some good scientific software out of it

sperm is stored in the female visceral fat, of course, everybody knows this

don't release preprints without a Methods section don't release preprints with a Supplementary Materials section don't release preprints without data don't release preprints without code use preprint servers to share your science, not advertisements

We quantified mRNA abundance, translation, protein abundance, protein degradation and cell growth across thousands of single cells from a mammalian tissue. The results revealed 𝐜𝐨𝐦𝐩𝐥𝐞𝐱 regulation & 𝐬𝐢𝐦𝐩𝐥𝐞 organizing principles: www.biorxiv.org/content/10.1... 🧵

There's been a recent explosion in shoddy, cookie-cutter analysis of large, widely-available public health datasets (think NHANES, the UK Biobank, FAERS, etc). Matt Spick and I offer some advice in our new Viewpoint in European Science Editing: doi.org/10.3897/ese.... More reading material below!

Fundamental Limits of Inferring Dynamical Gene Regulatory Models from Single-Cell Data https://www.biorxiv.org/content/10.1101/2025.09.12.674509v1

mega this, mega that. some of you computer people need to be more worried about charismatic MEGAfauna. i'm saying you're going to be eaten by tigers if you keep being confusing like this

Disappointed by the new iPhone that’s just been launched. No dry cleaning function or trouser press, limited cappuccino making, you can’t use it to raise the dead, and it has a weird metallic taste.

Large AI models are reported to achieve high accuracy (AUROC) predicting pathogenic variants across the genome. A preprint reports that the predictions are based on splice variants. Using only this info (no sequences, no AI) achieves AUROC=0.944 across noncoding variants. 1/2

Interesting work! And a perennial reminder that the deeper question "why should anyone want to do 2D visualization in the first place?" has no real answer.

Look at this crazy thing! www.nature.com/articles/s41...

🙃

They should have google maps but for searching things like "beautiful old oak tree near me", "cool, refreshing creek near me", etc. you see my vision

a great thing about using open-source software is when you want to use it in a way that seems obvious but is not documented and when you look into it not only is it impossible but the developer hates your specific use-case and built the entire application around not supporting it

Terrific stuff! the KB5060829 Windows update also broke casting from my Windows PC (which is about 90% of what I do with it), which is how you know Microsoft writes really good code and are well-positioned to help other people write good code with Copilot and so on