@marcrr.bsky.social
He/Him. Evolution and Bioinformatics. Posts mine alone, in English (mostly) & French. Chair of @dee-unil.bsky.social, Prof at @fbm-unil.bsky.social, group leader at @sib.swiss. PI of @bgee.org 🦣 Main account: https://ecoevo.social/@marcrr
Some thoughts on AI based on working on this project: Some thoughts on AI based on working on this project: liorpachter.wordpress.com/2026/02/19/t...
Preprints are living documents! All versions remain under "Info/History" with the same DOI. Revise from "Author Area" > "Submit a Revision", anytime pre-publication.
🔗 Check out our recently (Feb '26) updated bioRxiv preprint: doi.org/10.1101/833400 (2019)
#OpenScience #bioRxiv #medRxiv #openrxiv
Holy smoke. What ultimately happened???
In summary, we need to think much more carefully about how we benchmark these models, and also, there is still huge room for improvement. By and large, these models don't work nearly as well as most people think they do. 6/6
Key observation: If you compare prediction performance to a naive model that simply predicts the mean (known) fitness value at each site, pLMs barely win out, and on viral data they do worse! In other words, taking the mean value at each site does better than using the 650 million parameter pLM. 2/
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...
A. This is a horrific, preventable tragedy. B. Consistent rhetoric from authorities matters. I've held my 3 newborns & cried myself as their heels got poked & bled for their Vitamin K shots & the babies looked at me with Such Betrayal, but I did it b/c the AAP/FDA/Drs said it would keep them safe.
If any news outlets are covering this and want to talk to a trans Genomics PhD who specialized in human genetics and the evolution of sex, please reach out to me
This is complete nonsense scientifically. It's fascist eugenics, not biology
Setting a horrifying precedent for genocide
If 9 people stay on Twitter/X…
Posted by a friend on Instagram.
Yea, I see the resemblance to the other nine men. Taking it in seriously.
During #WomensHistoryMonth, I can personally reflect on this, and feel amazed at how women have made history despite this built in disadvantage.
Backwards, in heels, and harassed.
"How do you deal with microaggressions, hesitations and erroneous assumptions about your intrinsic worth as a colleague? Each incident is small, deniable and completely forgettable. But, cumulatively, they form a mighty river." #WomenSTEM #sexism
www.nature.com/articles/d41...
Les records tombent les uns après les autres. La France vient de vivre :
➡️la 1ère dizaine de mars la plus chaude jamais observée.
➡️la plus longue série discontinue au-dessus des normes avec maintenant deux mois entier.
➡️certaines floraison sont les plus précoces jamais observées.
1/3
With @saramitri.bsky.social, @sonjalehtinen.bsky.social and L. Lehmann we’ve launched the UNIL Center for Theory in Ecology and Evolution @unil.bsky.social🇨🇭
To kick things off, we’re offering short visiting fellowships for theorists in ecology & evolution. Apply & pls RP 😀
tinyurl.com/2wem36zz
Nouveau strip ! "Bilingue". Une histoire tragique sur fond de fossé linguistique.
À lire sur le blog: www.bouletcorp.com/rogatons/202...
Ou sur Insta (english version after the french one)
www.instagram.com/p/DVqnKkAiCj6/
Article publié : Foster, G., & Rahmstorf, S. (2026). Global warming has accelerated significantly. Geophysical Research Letters, 53, e2025GL118804.
6/6
Depuis 2015, le rythme du réchauffement a atteint environ 0,35 °C par décennie, contre environ 0,2 °C entre 1970 et 2015. Les données corrigées montrent qu’une accélération du réchauffement climatique est désormais détectable depuis 2015, avec une certitude statistique supérieure à 98 %.
2/6
J'ai toujours été prudent quant à l’idée d’une accélération du réchauffement mondial. À partir de maintenant, je ne le serai plus.
1/6
I live a sheltered life…
Last week I interacted for the 1st time directly with a real tech bro, the sexism was shocking in its blatancy. I'm used to biases and to hypocrisy, not to a guy defending biased AI developped by his startup male-only team as good.
The SMBE IDEA Task Force stands with all women doing science: cis and trans, in every country, under every condition. In war zones, under authoritarian regimes, where their existence is an act of resistance.
March 8 carries weight. Science is only as strong as who it includes.
#WomenInSTEM #March8
The US and Canada go to Daylight Savings Time tomorrow morning.
#OTD 10 years ago we published the spotted #gar genome. 600+ citations later, we are proud of its impact on vertebrate genomics. @jpostlethwait.bsky.social
Check out the new assemblies generated with @vertebrategenomes.bsky.social
www.ncbi.nlm.nih.gov/datasets/gen...
#EndlessFishMostBeautiful
There's a common misconception that zoonotic viruses require significant adaptation to jump from animals to cause human epidemics.
Not so 👇.
Further, we see clear signs of 1977 flu experiencing cell passage, prior to epidemic.
SARS-CoV-2? Business as usual.
www.sciencedirect.com/science/arti...
Now accepting nominations for the ESEB Underrepresented ECR Achievement Award!
⭐ Highlight the achievements of underrepresented ECRs
💰2 x €2000 awards available
🗓️ Deadline 1st JULY 2026
shorturl.at/YxepB
(OK le soutien financier était peanuts, on doit pouvoir améliorer ça.)
Les gars sur le point de réinventer le labo CNRS des années 1990
"a model of a research lab with a flatter hierarchy, individuals from different disciplines working together and stable long-term financial support that doesn’t rely on quick results or fast publication"
www.nature.com/articles/d41...
And to a lesser extent in male reproductive system…
There is a similar problem with branch-site dN/dS tests, and a solution has been to use the sum of ∆lnL over meaningful categories.
Here this sum over organs/tissues shows a strong excess of positive selection on regulatory sequences in human nervous system evolution 😎.
Update of our preprint on detecting selection on regulatory sequences! We notably added an analysis for human, where the likelihood test per ChIPseq peak lacks power.
www.biorxiv.org/content/10.1...
The best keynote talks make you wish you studied the question the speaker is presenting. In that respect, this amazing talk by Amita Sehgal (UPenn) on how and why flies sleep (which is likely the same reason we do) is just so amazing it’s hard to describe. Mind is truly blown. #Dros26
