Pierre-Paul Axisa

Is scientific reform an unwinnable arms race? Methodological improvements should, in theory, mean more robust evidence and inference, and more rapid advances in knowledge. However, these methods are often subsequently used in the pursuit of…

Improvements to methods should lead to more robust evidence and more rapid advances in knowledge. In their Perspective, Marcus Munafò & @mendelrandom.bsky.social look at why this hasn't been the case and what can be done about it #AcademicSky 🧪

a scene from "the breakfast club" with r package hexes superimposed on the five characters from left to right: the delinquent is a plain-text markdown table the jock is the janitor package the loner is the tinytable package the popular girl is the knitr package the nerd is the gt package the jock is the 'janitor' package, the

How it feels picking an #rstats package to make tables

A step in the right direction. But it’s not like reviewers/authors take those journal submission portal acknowledgements boxes very seriously anyways…

#bioinformatics #statistics #memes #bluesci

- Have you heard of Gaussian process?
- it rings a bell

I guess that is similar to ggrastr? I like ggrastr because you can rasterize a layer of choice

a picture of a monster with the words zanar aesthetics on the bottom right ALT: a picture of a monster with the words zanar aesthetics on the bottom right

Sauron called, he’s pissed that people are stealing his business model

I love the "Preferred sandwich making parent", definitely a thing with my 5yo.

I am disappointed there’s no analysis plan including the sandwich estimator 🥪

I wrote a blog for the Meta-Research Center expressing my infinite frustration about not getting data. What else is new, you might think? Well, I added an extra layer of annoyance directed at the journals who do NOTHING to enforce promised data sharing.

metaresearch.nl/blog/2026/2/...

That’s worries me the most, especially given the propensity of the bots to satisfy the user. If you prompt with an expected outcome, it might bend backwards to get you that outcome.

nebula: Negative Binomial Mixed Models Using Large-Sample Approximation for Differential Expression Analysis of ScRNA-Seq Data A fast negative binomial mixed model for conducting association analysis of multi-subject single-cell data. It can be used for identifying marker genes, differential expression and co-expression analy...

There’s the nebula package: cran.r-project.org/web/packages...

A group of Cambridge scientists stand together in a laboratory, smiling at the camera. One person holds a book titled ‘Self-Doubt’. A text banner reads: ‘Cambridge scientists share the truth behind their successes and reflect on the power of failure.’

What does it take to succeed in science? Failure.

Cambridge scientists from @labliston.bsky.social share the self-doubt behind their careers.

Self-Doubt by Adrian Liston from @stcatharines.bsky.social and the Department of Pathology is Out tomorrow 👇

https://bit.ly/4a4tWUE
#Careers #Science

I am deeply offended by boomerplyr (inspired by genzplyr), because I am NOT a boomer (squarely Xennial) and I "get" all the translations bradlindblad.github.io/boomerplyr/ #Rstats

Annotate(geom = "tile", x = inf, y = inf)

I’ve seen "expansion scores" that are just the log(#cells) for a given clone. But it sounds way more elaborate to say expansion score!

A tweet from Lior Pachter from July 26, 2024: "I think this paper is a Denial of Peer Review Attack (DOPRA). It's kind of like a DoS (denial of service) attack. There is so much data, so many methods, so much code, so many figures, so many panels, so much supplement, so much text, that it is overwhelming. 18/"

every big data paper with a bold story that is impossible to comprehend, evaluate, and independently verify reminds me of DOPRA. i've come to increasingly appreciate small, unassuming papers with humble conclusions that you can track word for word, data point by data point, assumption by assumption.

Wasn’t cell reports meant to be that?

New preprint alert: we use sign errors as a test of how well TWAS works.

Very worryingly we find that TWAS gets the sign wrong around 1/3 of the time (compared to 50% for pure guessing). You can read more about our analysis here, and what we think is going on 👇

Most researchers would receive more recognition if assessed by article-level metrics than by journal-level metrics Are authors fairly judged by assessment of the prestige of the journals in which their work is published? This study compares article level metrics with journal level metrics, finding that the vast ma...

"We also find that approximately half of researchers never publish in a venue with an impact factor above 15, which, under journal-level evaluation regimes, may exclude them from consideration for opportunities. Many of these researchers publish equally influential work"

An instruction sheet telling lab users to warn security in case of various emergencies. Someone has written in "zombies" to that list.

Remember: only YOU can prevent the zombie apocalypse

Sorry I cannot join the faculty meeting today, I need to urgently fill out the war outbreak report form. Yet another case of scientists swamped by bureaucracy 😆

Twins Are So Much More Interesting Than Heritability Estimates On starting places, "missing environmentality," and the Waddington landscape of life

I wrote about missing heritability, "missing environmentality," and why I still think twin studies are interesting and valuable: kathrynpaigeharden.substack.com/p/twins-are-...

Massive single-cell study by Kanai et al (www.medrxiv.org/content/10.1...):
- Once statistical power is high, constrained genes have more (though weaker) eQTLs.
- Chromatin-QTLs near constrained genes have "normal" effect sizes, colocalize more with disease, but exhibit attenuated peak-gene effects.

A thread of directed acyclic graphs (DAGs) that look like record covers... because that's EXACTLY what the world needs

1. Huey Lewis and the News: link.springer.com/chapter/10.1...

Well, I found the solution Harmit!

Americans make great biologists because we are forged by the hardest challenge in all of biotech: getting flavor from a turkey

You covered previously the recent family GWAS from Tan et al.. How does that relate conceptually to the other approaches? It’s similar to RDR? But it only uses common variants so we miss the rare variants’ contributions?

Why we made Affinity free, and how we’ll keep it that way We’ve made Affinity completely free, empowering professional designers with studio-grade creative software, supported by Canva’s sustainable ecosystem.

I'm a longtime fan of Affinity Designer as an affordable Illustrator-killer for figures, and... it's now free?! www.canva.com/newsroom/new...
Highly recommended if you're sick of paying Adobe $. Maybe Canva can buy NPG too and get rid of the OA fees.