iiinova — where your digital self keeps evolving

Reposted by Jérôme Arnoux

IceLab @icelabumu.bsky.social · 23/09/2026

Transcriptomics, proteomics, computational modelling and experiments. The work includes researchers across IceLab, @mimsumea.bsky.social, UCMR and @scilifelab.se — including Sena Gizem Süer, @jpjarnoux.bsky.social, Cemal Erdem, @andrenmateus.bsky.social and @kavican.bsky.social

121

Jérôme Arnoux @jpjarnoux.bsky.social · 17/09/2026

🙏 Co-authored with Sena Gizem Süer. Thanks to @kavican.bsky.social for supervising the project, Cemal Erdem and André Mateus for their advice, @icelabumu.bsky.social for the fertile environment, and the whole team.

000

Jérôme Arnoux @jpjarnoux.bsky.social · 17/09/2026

💻 Beyond the biology, this gives the field a paired mRNA–protein dataset across multiple pathogens and stress conditions, exactly the kind of data needed to train machine learning models (what I'm working on now in my postdoc).

100

Jérôme Arnoux @jpjarnoux.bsky.social · 17/09/2026

🔬 Why? We traced it to impaired translation initiation for Yersinia pseudotuberculosis, confirmed not just computationally but experimentally too. Under osmotic stress, bacteria keep transcribing, but struggle to actually build the proteins.

100

Jérôme Arnoux @jpjarnoux.bsky.social · 17/09/2026

🧂 One condition stood out: osmotic stress. Across all three pathogens, mRNA–protein correlation dropped the most here for stimulon genes, more than for temperature, acid, or nutrient stress.

100

Jérôme Arnoux @jpjarnoux.bsky.social · 17/09/2026

📉 mRNA is often used as a stand-in for protein levels, but we found this proxy weakens specifically for genes with the strongest transcriptional and translational stress response. The bigger the stress signal, the less you can trust it.

100

Jérôme Arnoux @jpjarnoux.bsky.social · 17/09/2026

🦠 Bacterial gene expression is not a reliable proxy for protein levels, at least not under stress. Our new paper in PNAS shows how far mRNA and protein diverge, across three pathogens and 10 stress conditions. www.pnas.org/doi/10.1073/... #Pathogen #RNA #protein #MachineLearning

pnas.org

PNAS

Proceedings of the National Academy of Sciences (PNAS), a peer reviewed journal of the National Academy of Sciences (NAS) - an authoritative source of high-impact, original research that broadly spans...

131

Jérôme Arnoux @jpjarnoux.bsky.social · 17/09/2026

🔬Why? We traced it to impaired translation initiation for Yersinia pseudotuberculosis, confirmed not just computationally but experimentally too. Under osmotic stress, bacteria keep transcribing, but struggle to actually build the proteins.

000

Jérôme Arnoux @jpjarnoux.bsky.social · 17/09/2026

🧂 One condition stood out: osmotic stress. Across all three pathogens, mRNA–protein correlation dropped the most here for stimulon genes, more than for temperature, acid, or nutrient stress.

100

Jérôme Arnoux @jpjarnoux.bsky.social · 17/09/2026

100

Reposted by Jérôme Arnoux

David Vallenet @vallenet.bsky.social · 10/08/2026

🧬 PanGBank is now available! • 4,600+ prokaryotic #pangenomes, 393,000+ genomes • #PPanGGOLiN graph partitioning, RGPs & Modules • User-friendly web interface, API & CLI 🌐 pangbank.genoscope.cns.fr 🎓 Try it in on #AMR in A. baumannii : shorturl.at/Vw3zc #ABRomics @ifb-elixir-fr.bsky.social

273

Jérôme Arnoux @jpjarnoux.bsky.social · 20/07/2026

8/ Co-author Jean Mainguy, Laura Bry, Quentin Fernandez De Grado, Yazid Hoblos, @vallenet.bsky.social, Alexandra Calteau @genolabgem.bsky.social

001

Jérôme Arnoux @jpjarnoux.bsky.social · 20/07/2026

7/ Application We scaled this up to 6,000+ genomes across 4 Enterobacteriaceae species, showing PANORAMA can annotate, compare, and explore diversity at a scale single-genome methods can't reach.

101

Jérôme Arnoux @jpjarnoux.bsky.social · 20/07/2026

6/ Comparison PANORAMA is the first tool including a pangenome comparison. Based on pangenome feature predicted, as systems or spot of insertion, PANORAMA identify shared feature, revealing potential common evolutionary mechanisms across species.

100

Jérôme Arnoux @jpjarnoux.bsky.social · 20/07/2026

5/ Benchmark We validated this on 941 Pseudomonas aeruginosa genomes, benchmarking against reference tools PADLOC (github.com/padlocbio/pa...) and DefenseFinder (github.com/mdmparis/def...). PANORAMA predictions are highly similar to references while being exponentially more efficient.

100

Jérôme Arnoux @jpjarnoux.bsky.social · 20/07/2026

4/Sytems detection PANORAMA uses a novel 3-level rule-based model to describe systems, then runs a genomic context search based on that model at the pangenome level — finding systems without needing single-genome scans. Model collection is available here: github.com/PANORAMA-mod...

100

Jérôme Arnoux @jpjarnoux.bsky.social · 20/07/2026

3/Architecture PANORAMA builds on PPanGGOLiN pangenome graphs (nodes = gene families, edges = genomic neighborhood) to do two things: 1) detect biological systems directly in the graph 2) compare conserved features across pangenomes, like biological systems or spot of insertion.

100

Jérôme Arnoux @jpjarnoux.bsky.social · 20/07/2026

2/Problem Genome sequencing has exploded, and analyzing genomes one by one doesn't scale. Pangenomes give a better picture of diversity, evolution, and adaptation. Until now, no tool exists to predict and compare bilogical systems at the pangenome level. PANORAMA fill this gap.

100

Jérôme Arnoux @jpjarnoux.bsky.social · 20/07/2026

1/ Summary We built PANORAMA, a tool to predict and compare biological systems (like phage defense) across thousands of bacterial genomes at once, using pangenome graphs. Built on PPanGGOLiN (github.com/labgem/PPanG...), PANORAMA has been benchmarked on 900+ genomes and scaled to 6000+.

100

Jérôme Arnoux @jpjarnoux.bsky.social · 20/07/2026

Proud to announce that PANORAMA just came out in @plos.org Computational Biology🧬 Paper: journals.plos.org/ploscompbiol... Code: github.com/labgem/PANOR... Docs: panorama.readthedocs.io/latest/ #Bioinformatics #Pangenomics #ComparativeGenomics #Microbiology #OpenScience 🧵 Here's what it does 👇

152

Reposted by Jérôme Arnoux

LABGeM @genolabgem.bsky.social · 15/07/2026

🧬 We're thrilled that PANORAMA has been published! An open-source method to predict & compare biological systems across bacterial pangenomes. Built on PPanGGOLiN, it runs at #pangenome scale in minutes vs. hours for genome-by-genome tools. 📄 doi.org/10.1371/jour... @jpjarnoux.bsky.social

045

Reposted by Jérôme Arnoux

CNRS @cnrs.fr · 03/03/2026

En élucidant le mécanisme qui rend le champignon microscopique Candida albicans capable de provoquer des mycoses, une équipe du CNRS ouvre la voie à de nouvelles pistes thérapeutiques.

lejournal.cnrs.fr

Comment une levure devient mycose

01310

Reposted by Jérôme Arnoux

IceLab @icelabumu.bsky.social · 06/02/2026

New tutorial on the map equation from IceLab's @mrosvall.bsky.social 👍 see post below!

041

Reposted by Jérôme Arnoux

CNRS @cnrs.fr · 04/02/2026

#Communiqué 🗞️ Une équipe a cartographié les correspondances entre les sillons cérébraux et les empreintes laissées sur les endocrânes à partir de sujets vivants afin de mieux reconstituer les cerveaux de nos ancêtres. 🧠 👉 www.cnrs.fr/fr/presse/le...

0147

Jérôme Arnoux @jpjarnoux.bsky.social · 02/02/2026

Happy to announce that I've started my postdoctoral position at @icelabumu.bsky.social ! 🎉 My project focuses on developing AI models (ML/DL) to predict protein levels from RNA-seq data in bacterial infections, to identify new therapeutic targets to combat #AMR🦠🤖 #AIforScience #Postdoc

031

Reposted by Jérôme Arnoux

LABGeM @genolabgem.bsky.social · 12/06/2025

Congratulations to @jpjarnoux.bsky.social who just achieved a major milestone by successfully defending his thesis! 🎓✨ Welcome to the world of doctors! #AcademicLife #ProudPI #pangenomics @genoscopeumr.bsky.social‬ @univparissaclay.bsky.social @cea.fr @univevry.bsky.social

022