Fantastic new paper casting doubt on explainability of explainable AI. To explain complex machine learning algorithms you need reproducibility of the explanation at a minimum academic.oup.com/ehjdh/articl... #machinelearning #Statistics #StatsSky @maartenvsmeden.bsky.social

The practice of trial and error generates ingenuity. A fresh-faced naive student, or intern, is the ultimate idea generator. A collaborator. A creator. A chance to take an approach no one's thought to take before - not even you. Be selfish. Train others. Do more than you alone can imagine. 13/13

Associations and Interaction Effects of Socioeconomic, #Lifestyle, and Genetic Factors on Intrinsic Capacity Evidence suggesting interactions between #genes and #environment in the definition of IC #Geriatrics #Genetics #Ageing #Health #Science #MedMastodon academic.oup.com/biomedgeront...

🧪 #genetics #alzheimer's #parkinson's Interested in the genetics and molecular biology of Alzheimer's and/or Parkinson's? Stellar effort led by Yann Le Guen and Andres Peña-Tauber: www.medrxiv.org/content/10.6... Largest-to-date gene burden studies in AD & PD. Powerful data browser (link in paper).

How can #genetics transform drug discovery? In his #NAS163 Research Briefing, David Altshuler (Vertex Pharmaceuticals, Inc.) will discuss how linking genetic variation to disease biology helps scientists identify therapeutic targets and treatments. Register: nasonline.swoogo.com/NAS163_resea...

In a new study, Sarah Marion &co use genetic mapping and whole genome sequencing in wild #Drosophila to reveal the genetic basis of naturally occurring lethal mutations and introduce a new model that could explain their frequencies in natural populations. 🧪 #genetics

When AI tools do well - or badly - at a task, it’s tempting to explain it away with the benefit of hindsight. But can you anticipate how well they’ll do before you see the outcome? Have a go here – and then see how your estimates compare with 1000+ others: quiz.wholesum.tech

Theorem of the Day (March 15, 2026) : Praeger’s Theorem on Bounded Movement Source : Theorem of the Day / Robin Whitty pdf : buff.ly/OjxKS72 notes : buff.ly/tL31ux6 #mathematics #maths #math #theorem

A beloved Oxford cinema is under threat from college expansion plans. Anyone who’s been will know how special it is. Join 10,000 supporters and sign the petition. www.change.org/p/save-the-u...

It used to be that only at most three or maybe four people read a PhD. The author, the supervisor, the examiner(s). Nowadays even the “author” might not have read it.

This is a fun quiz! I’ll put my results in a reply since they contain spoilers for the answers

Asked to speculate about failure to replicate, the students come up with various creative ideas. But it's rare that anyone suggests the big cool findings of the original study may simply have been false positives i.e. the claimed effects weren't real in the first place. Some find it shocking😲...2/n

A fair coin "knows" π. Stop flipping when heads first outnumber tails. The expected fraction of heads is π/4. Average many trials → estimate π. Result by J. Propp, animated for #PiDay. Made with #Python #NumPy #Matplotlib and @marimo.io #Mathematics #MathArt #CreativeCoding

🧬New today! 📄Inclusion bias affects common variant discovery and replication in a health-system linked biobank 🧑‍🤝‍🧑 @loldeloo.bsky.social @apimplaskar.bsky.social @bpasaniuc.bsky.social & colleagues

Many journals require data and code sharing. Yet code is still rarely shared and datasets are often hard to reuse. Our new paper introduces the @sortee.bsky.social guidelines for data & code quality control in ecology & evolution, developed by 26 experienced data editors. 📄 doi.org/10.24072/pcj...

I wish more people knew this. Power analysis should be based on the smallest effect size of interest. Not on a guess, or a hope. You also need to specify that effect to make your claim falsifiable, and to know when the effect is statistically significant, but practically irrelevant.

“SVM, NN and RF may need over 10 times as many events per variable to achieve a stable AUC and a small optimism than classical modelling techniques such as LR. This implies that such modern techniques should only be used in medical prediction problems if very large data sets are available.“ #stats

Happy #PiDay! Here's the first 1,000 digits of pi encoded as colours 🎨 made with #RStats

@hggadvances.bsky.social's latest article introduces practical approaches for leveraging multiple observations to assess impacts on GWAS power, SNP-heritability, gene set enrichment, and polygenic prediction: https://www.cell.com/hgg-advances/fulltext/S2666-2477(26)00026-6 #ASHG #HumanGenetics

Causal Genotype–Phenotype Relationships from Population-Sampled Parent–Child Trios: TMT and TDT onlinelibrary.wiley.com/doi/10.1002/... #statgen

Balancing selection alert!! 🧬🧪 New preprint where we try to quantify how likely (or rather, unlikely) it is for balancing selection to maintain stable polymorphism, and how easy (or rather, challenging) it is to identify its signatures in genomes. #Popgen #MolecularEvolution #EvoBio #Science

Another day, another stupid Excel chart.

If we're going to solve this we need to have an honest and informed conversation about the true costs of doing things 2/2 journals.plos.org/plosbiology/...

Dogs evolved to eat your leftovers! Comparing dog & wolf genomes revealed dogs have up to 30 EXTRA copies of the amylase gene (AMY2B) that helps digest starch. This is a key genomic signature of living alongside humans & table scraps for thousands of years 🐕 www.nature.com/articles/nat... #2026MMM

Oft Herr Mendel dreams that genes form living matter. They do the latter. doi.org/10.1093/gene...

New review on context-specificity and disease genetics from @ygilad.bsky.social and me: authors.elsevier.com/sd/article/S...

A new paper on anomaly detection: robjhyndman.com/publications... The methods are implemented in the weird #rstats package.

When Referee 3 is too lazy to review the paper and their AI-generated response cites hallucinated papers to “prove” non-originality! This is mind-boggling. #AcademicSky

Still manually merging UK Biobank-linked health data? Faffing with Swiss Army Knife to extract variants or create polygenic scores? 🧬 I built {ukbrapR} so we could spend more time on science and less on data plumbing If it helps you, please cite! 📊 citation('ukbrapR') github.com/lcpilling/uk...

New paper showing that much of the apparent success of protein language models in predicting mutational effects is a mirage: These models mostly memorize sites. 1/ www.biorxiv.org/content/10.6...

What do you mean I have to publish another paper after I am done with this one. That doesn't sound right. That sounds like a pyramid scheme

If you're looking to build or deepen your knowledge in statistical genetics, the ISG Workshop (June 1–11) covers the full range: biometrics, GWAS, polygenic scores, causal inference, and more. Open to all levels, virtual, and international: www.colorado.edu/ibg/workshop... Retweets appreciated 🙏

Yes, academic publishing is broken. But these heroes have a plan to fix it... https://arxiv.org/abs/2304.01393 #academicchatter

Last year Leonardo Ripoli and I archived the paper "Improved MCMC with active subspaces". We identify directions in the parameter space in which the likelihood is not informative, and explore ways of making MCMC more efficient by focussing on the directions where it is. arxiv.org/abs/2501.0...

“Where code has been shared in a non-permanent repository, such as Github or Bitbucket, a snapshot of the code should be archived in a permanent, public repository, for example, Zenodo, Code Ocean, or the Software Heritage archive.” #OpenScience #code #CodeSharing

In somebody else's words:

If you use #QuartoPub for any PDF documents for teaching and you need to meet impending PDF accessibility rules, upgrade to v1.9, add format: typst: pdf-standard: ua-1 to the YAML front matter, and it'll work!

📣 SING USA 2026 applications are open! The Summer Internship for Indigenous Peoples in Genomics (SING) workshop will be July 19–24, 2026 at UW–Madison. ⏰ Apply by March 13 (midnight Hawaiʻi time) Learn more & apply: sing.nativebio.net Please share with Indigenous students! #Indigenous #Genomics

Since it's not on @biorxivpreprint.bsky.social, my normal crowd might not see it, but I'm very excited about this recent work that, actually, started as class project in my graduate genomics class last year: eprint.iacr.org/2026/463 . Zero knowledge proofs for genomics; with algorithms :).

In case anyone needs it for their syllabi, my statement in gen-Ai from the minicomic I made as a syllabus for class last semester. All online and printable here spinweaveandcut.com/fall-2025-sy...

In more important spite-driven development news... sessioncheck is now on CRAN. At long last I now have something dead-simple and concrete I can point to next time I'm asked to put rm(list=ls()) in a script for no obvious reason sessioncheck.djnavarro.net

The data is in: the NIH goalposts have shifted. What were once almost certain fundable scores have become coin flips and what used to be likely grants have become aspirational, leading to fewer awards. Another manifestation of how HHS policies have led to fewer awards and less science.

Excited to share this preprint that describes my latest work on using GPUs to accelerate processing of RNA-seq data. The title says it all: "RNA-seq analysis in seconds using GPUs" now on biorxiv www.biorxiv.org/content/10.6... and github github.com/pachterlab/k... Figure 1 shows they key result

I can't wait to try this! I'm always futzing around saving plots to different dimensions and aspect ratios, then opening the files to see if they're ok, then forgetting to close them and getting errors because the filename is in use when I try to re-save with different parameteres.

These captchas just keep getting harder #rstats

ggview::canvas() is the best thing I have learned in 2025!

This is awesome! No more (change code -> save plot -> open file)^n just to make sure your plot is readable

The canvas() function from the ggview R package is very useful for previewing/tweaking a ggplot into publication-ready format: it renders a plot "as it would appear if saved to a file with the specified dimensions". cran.r-project.org/web/packages... #RStats #ggplot #ggview

Some work freshly published at EJS: projecteuclid.org/journals/ele... 'Analysis of Multiple-try Metropolis via Poincaré inequalities' - Rocco Caprio, Sam Power, Andi Q. Wang We conduct a convergence analysis of a specific class of MCMC procedures based on multiple-proposal strategies.