Sign in

Constantin Ahlmann-Eltze

@const-ae.bsky.social
2.1K followers 694 following 70 posts

Scientist at Isomorphic Labs. Previously at UCL and EMBL working with Wolfgang Huber. Biology, ML, Python, R, cancer immunology

PostsRepliesMedia
Reposted by Constantin Ahlmann-Eltze
Emma Dann @emmamarydann.bsky.social · 28/08/2026
Our work on systematic perturb-seq of primary human T cells is now out in Cell 🎉 www.cell.com/cell/fulltex... It's been a privilege to work with @ronghuizhu.bsky.social between @jkpritch.bsky.social @marsonlab.bsky.social labs, with a dream-team of co-authors ❤️ Highlights in preprint thread👇
cell.com
Genome-scale perturb-seq in primary human CD4+ T cells maps context-specific regulators of T cell programs and human immune traits
A dynamic atlas of gene regulation was generated by perturbing every expressed gene across 22 million primary human CD4+ T cells under resting conditions and following re-stimulation. The resulting ma...
14620
Reposted by Constantin Ahlmann-Eltze
Richard Sever @richardsever.bsky.social · 06/07/2026
The other thing I feel needs continued emphasis is pretty much all these preprints are written with the expectation they will undergo peer review. That may reduce the tendency of authors to overclaim from the outset and it tells us little about what they would write without prospect of formal review
2162
Reposted by Constantin Ahlmann-Eltze
Wolfgang Huber @wkhuber.bsky.social · 01/07/2026
EMBL's new AI unit is recruiting. We are seeking a machine learning scientist to establish their independent research group bridging innovations in machine learning/AI with modern biology. embl.wd103.myworkdayjobs.com/en-US/EMBL/j...
23226
Reposted by Constantin Ahlmann-Eltze
Luke Zappia @lazappi.bsky.social · 10/06/2026
Excited that the {anndataR} 📦 publication is out academic.oup.com/bioinformati... 🎉! I hope this makes it easier to move single-cell data between R and Python and encourages people to take advantage of both languages. Check it out on @bioconductor.bsky.social bioconductor.org/packages/ann....
13011
Constantin Ahlmann-Eltze @const-ae.bsky.social · 27/05/2026
Just wanted to say that I appreciate the live posting :)
000
Constantin Ahlmann-Eltze @const-ae.bsky.social · 27/05/2026
Yes, I was surprised when I saw that. The good thing is that CRAN isn’t the only option anymore :)
010
Constantin Ahlmann-Eltze @const-ae.bsky.social · 27/05/2026
Thank you for the thoughtfully designed package :)
100
Constantin Ahlmann-Eltze @const-ae.bsky.social · 26/05/2026
I have been working with polars a bit lately and was missing some of my favorite functions from {{forcats}} and {{santoku}} for dealing with enums/factors. So, naturally, I wrote a new package to fix this :) const-ae.github.io/polarstation/ #polars #python #rstats
const-ae.github.io
polarstation
1101
Reposted by Constantin Ahlmann-Eltze
Grant McDermott @gmcd.bsky.social · 29/04/2026
Happy to see that `jgd` is now up on CRAN. An alternative #rstats graphics device that's fast and lightweight (mostly pure C). Try it out with our VS Code extension or Deno server. More details: github.com/grantmcdermo... Joint work with the indefatigable @eitsupi.bsky.social.
2396
Reposted by Constantin Ahlmann-Eltze
Valentine Svensson @nxn.se · 22/04/2026
I don't think cell representation foundation models _should_ be good at zero-shot integrating data. How do you know which variation I want to integrate out?
052
Reposted by Constantin Ahlmann-Eltze
Constantin Ahlmann-Eltze @const-ae.bsky.social · 22/03/2026
Thrilled to join Isomorphic Labs—time to build better ML models of cells and create the drugs of the future!! 🧬🧫🧑‍💻🥳
Picture of myself in front of the Iso logo
2241
Constantin Ahlmann-Eltze @const-ae.bsky.social · 22/03/2026
Thrilled to join Isomorphic Labs—time to build better ML models of cells and create the drugs of the future!! 🧬🧫🧑‍💻🥳
Picture of myself in front of the Iso logo
2241
Reposted by Constantin Ahlmann-Eltze
VIB.AI @vibai.bsky.social · 18/03/2026
We are super excited to welcome Dmitry Kobak @hippopedoid.bsky.social as a new VIB.AI group leader! The Kobak lab will officially kick off in April at the VIB.AI Ghent hub. More: vib.ai/en/news/dmit...
0175
Reposted by Constantin Ahlmann-Eltze
Wolfgang Huber @wkhuber.bsky.social · 15/03/2026
1/ Apply for an inter-disciplinary postdoc position in Heidelberg with the Health & Life Science Alliance Deadline: **31 March** Project proposals can be self-designed or adapt one of the outlines on the website ("List of project outlines for download") www.health-life-sciences.de/opportunitie...
health-life-sciences.de
HEALTH + LIFE SCIENCE ALLIANCE | Interinstitutional Postdoc Call |…
Apply for the Alliance Interinstitutional Postdoc Program in health & life sciences. Two-year funded positions with interdisciplinary research support.
11218
Constantin Ahlmann-Eltze @const-ae.bsky.social · 05/03/2026
I have also seen, from people who know base really well, incredibly elegant code using tapply, merge, and clever indexing. But, in my experience, the typical code of a medium experienced person that uses the tidyverse is more robust than the one that uses base.
210
Constantin Ahlmann-Eltze @const-ae.bsky.social · 05/03/2026
It's an example. As I said, the tidyverse has the advantage that it provides functions for typical data analysis tasks all in one place with great documentation and common semantics. Also, personally I'm not a fan that merge auto-sorts the output. Another example would be pivot_longer :)
100
Constantin Ahlmann-Eltze @const-ae.bsky.social · 05/03/2026
Teach the tidyverse! I have seen the code from people, who only learned base, that reinvents a join function and it's not pretty 😅 In the end, the tidyverse is a collection of useful functions for typicy data analysis tasks and beginners benefit from being told those exist :)
110
Reposted by Constantin Ahlmann-Eltze
Sjors Scheres @sjorsscheres.bsky.social · 16/10/2025
I just made an account and voted for this biology laboratory made of LEGO to "promote biological research and inspire more people into the world of biology". They need 10K votes. Reposting and/or voting below would be helpful. 🤗 beta.ideas.lego.com/product-idea...
beta.ideas.lego.com
Biological Laboratory | LEGO® Ideas
This is a two-floor realistic Biological Laboratory building. In this design, I try to tell stories about biologists, introduce biology concepts to general publ…
8223131
Reposted by Constantin Ahlmann-Eltze
Mark O’Neill 🐻 @radiobeartime.com · 22/02/2026
www.deadlanguagesociety.com/p/how-far-ba... I cannot speak as to the accuracy of the piece yet I found it a fascinating exercise @adamroberts.bsky.social @drlauravarnam.bsky.social
deadlanguagesociety.com
How far back in time can you understand English?
An experiment in language change
2212064
Reposted by Constantin Ahlmann-Eltze
Davis Vaughan @davisvaughan.bsky.social · 16/02/2026
screaming this from the rooftops
1142
Reposted by Constantin Ahlmann-Eltze
Wolfgang Huber @wkhuber.bsky.social · 13/02/2026
Job offer: PhD-level biostatistician with ML&omics experience to work on prediction models for CAR T cell therapy, in an exciting clinical trial - spatial omics - data science consortium (LOOP/INTeRCePT3.0 centered in Zurich; this position in Heidelberg) embl.wd103.myworkdayjobs.com/en-GB/EMBL/j...
Screenshot of Job AdEMBL site in HeidelbergZurich, incl. a bit of UZH/USZ
0115
Constantin Ahlmann-Eltze @const-ae.bsky.social · 31/01/2026
It definitely helps, but I think it's also simply that most scientists have a LinkedIn account and occasionally check it, whereas still only a minority has a bluesky account.
000
Constantin Ahlmann-Eltze @const-ae.bsky.social · 31/01/2026
Yeah, the amount of engagement on LinkedIn compared to bluesky is crazy. I recently posted about a new paper and it got 480 likes on LinkedIn vs 7 on bluesky (and I have more followers here than there...) :/
120
Constantin Ahlmann-Eltze @const-ae.bsky.social · 21/01/2026
We wrote a review of representation learning methods of single-cell RNA-seq data, where we compare factor models, autoencoders, contrastive learning, and foundation models 🎉 rnajournal.cshlp.org/content/earl...
rnajournal.cshlp.org
Representation learning of single-cell RNA-seq data
A monthly journal publishing high-quality, peer-reviewed research on all topics related to RNA and its metabolism in all organisms
182
Reposted by Constantin Ahlmann-Eltze
Valentina Boeva @valboeva.bsky.social · 20/01/2026
Boeva Lab is #hiring postdocs and senior scientists in AI/ML for Cancer Biology! Check the offer and apply at: jobs.ethz.ch/job/view/JOP...
jobs.ethz.ch
Postdoctoral Researcher or Senior Scientist (AI × Biology)
088
Reposted by Constantin Ahlmann-Eltze
Wolfgang Huber @wkhuber.bsky.social · 15/01/2026
Apply for the Ascona workshop "Statistical and AI methods for multi-modal multi-scale modeling of biological systems", 28 Jun-3 Jul 2026 on Monte Verità, Lago Maggiore at the foot of the Swiss Alps. ascona2026.sciencesconf.org
View from the hotel roomPoster session 2024, with Valentina Boeva, Constantin Ahlmann-Eltze and othersWednesday afternoon hike incl. swim in the mountain riverAnother view from the hotel room
02215
Reposted by Constantin Ahlmann-Eltze
christofseiler.bsky.social @christofseiler.bsky.social · 14/01/2026
Save the date: April 9 from 4pm to 6pm CET. Our department is hosting an online seminar with @noeliaferruz.bsky.social @sdomcke.bsky.social @const-ae.bsky.social who will talk about models for protein design, large-scale perturbation screens, and benchmarking of perturbation prediction models.
063
Constantin Ahlmann-Eltze @const-ae.bsky.social · 02/12/2025
Join me in 5.5h, when we discuss our benchmark of perturbation prediction models and what the right metric is to assess if a gene expression prediction is good!
070
Reposted by Constantin Ahlmann-Eltze
Junyan Lu @junyanlu.bsky.social · 07/10/2025
🚀 Excited to share our new preprint: msBayesImpute - A Versatile Framework for Addressing Missing Values in Biomedical Mass Spectrometry Proteomics Data 👉 Improves imputation accuracy, normalization, and differential expression detection 📝https://www.biorxiv.org/content/10.1101/2025.10.02.679746v1
1162
Reposted by Constantin Ahlmann-Eltze
James Bonfield @jbonfield.bsky.social · 15/09/2025
Heads up: ignore samtools dot org, similarly minimap2 dot com and likely others. It's owned by a known phishing site and while the binaries they offer look valid currently (but note they may be serving us different binaries to others), that could change. Ie: it's not us (Samtools team)! Be warned
2145126
Reposted by Constantin Ahlmann-Eltze
Louise Deconinck @louisedck.bsky.social · 25/08/2025
We're excited to share that our preprint on anndataR, a new package bringing Python's AnnData to R, is now available on bioRxiv 🎉 🔗 Read the paper: www.biorxiv.org/content/10.1... 💻 Check the package in action: anndatar.data-intuitive.com
2257
Reposted by Constantin Ahlmann-Eltze
W. Joel Schneider @w-joel-schneider.bsky.social · 20/08/2025
Now on CRAN, ggdiagram is a #ggplot2 extension that draws diagrams programmatically in #Rstats. Allows for precise control in how objects, labels, and equations are placed in relation to each other. wjschne.github.io/ggdiagram/ar...
An arrow with a LaTeX equationTrigonometric functions and a unit circleA bivariate change model with structured residualsA hierarchical model of cognitive abilities
1018074
Constantin Ahlmann-Eltze @const-ae.bsky.social · 10/08/2025
Makes sense. I imagine this would simply be my primary use case, and I would prefer not having to refer to `x` twice. Something like: replace_values <- function(x, ..., from=NULL, to=NULL, lookup=NULL){ if(! is.null(lookup)){ from <- names(lookup) to <- lookup } ... }
100
Constantin Ahlmann-Eltze @const-ae.bsky.social · 10/08/2025
This looks great! I think it would be neat if the functions also supported named look-up vectors in addition to the `to` and `from` arguments :)
100
Reposted by Constantin Ahlmann-Eltze
Constantin Ahlmann-Eltze @const-ae.bsky.social · 04/08/2025
Our paper benchmarking foundation models for perturbation effect prediction is finally published 🎉🥳🎉 www.nature.com/articles/s41... We show that none of the available* models outperform simple linear baselines. Since the original preprint, we added more methods, metrics, and prettier figures! 🧵
Beeswarm plot of the prediction error across different methods of double perturbations showing that all methods (scGPT, scFoundation, UCE, scBERT, Geneformer, GEARS, and CPA) perform worse than the additive baseline.Line plot of the true positive rate against the false discovery proportion showing that none of the methods is better at finding non additive interactions than simply predicting no change.
212557
Reposted by Constantin Ahlmann-Eltze
James Reading @jamesreading.bsky.social · 05/08/2025
🚨 PhD Position available in our lab 🚨 exploring the power of blood immune multi-omics to detect lung cancer years prior to clinical diagnosis in a unique cohort of >10,000 CT screened individuals. ✅ Wet & dry lab ✅ September 2025 enrolment ✅ UK tuition fees only www.ucl.ac.uk/medical-scie...
ucl.ac.uk
Pre-Cancer Immunology
The Pre-Cancer Immunology Lab (James Reading Lab) is mapping pre-invasive T cell dynamics during carcinogenesis to detect and intercept cancer development.
01615
Reposted by Constantin Ahlmann-Eltze
Carl Zimmer @carlzimmer.com · 04/08/2025
I wrote about AI foundation models for biology last year: www.nytimes.com/2024/03/10/s... Benchmarking tests since then aren't finding that they're better than simpler models of how genes and cells work.
nytimes.com
A.I. Is Learning What It Means to Be Alive (Published 2024)
13712
Reposted by Constantin Ahlmann-Eltze
Nature Methods @natmethods.nature.com · 04/08/2025
An analysis shows that current deep learning models do not beat linear baselines in predicting gene perturbation effects, thus emphasizing the importance of further method development and evaluation. @const-ae.bsky.social @wkhuber.bsky.social @s-anders.bsky.social www.nature.com/articles/s41...
nature.com
Deep-learning-based gene perturbation effect prediction does not yet outperform simple linear baselines - Nature Methods
The analysis presented in this Brief Communication shows that, despite their complexity, current deep learning models do not outperform linear baselines in predicting gene perturbation effects, thus e...
06425
Constantin Ahlmann-Eltze @const-ae.bsky.social · 04/08/2025
Haha, would also be a succinct summary of most of my academic work 😅
130
Constantin Ahlmann-Eltze @const-ae.bsky.social · 04/08/2025
And lastly, a big shout-out to @wkhuber.bsky.social and @s-anders.bsky.social! Link to pdf: rdcu.be/ey7x0
rdcu.be
Deep-learning-based gene perturbation effect prediction does not yet outperform simple linear baselines
Nature Methods - The analysis presented in this Brief Communication shows that, despite their complexity, current deep learning models do not outperform linear baselines in predicting gene...
070
Constantin Ahlmann-Eltze @const-ae.bsky.social · 04/08/2025
- Li et al. doi.org/10.1101/2024.12.23.630036 - Chen Li et al. doi.org/10.1101/2024.12.20.629581 - Wong et al. doi.org/10.1093/bioinformatics/btaf317 And probably many more that I am missing here.
1100
Constantin Ahlmann-Eltze @const-ae.bsky.social · 04/08/2025
- @kasparmartens.bsky.social et al. openreview.net/forum?id=eb3ndUlkt4 - Gaudelet et al. doi.org/10.48550/arXiv.2404.16907 - @aaronwtr.bsky.social et al. openreview.net/forum?id=t04D9bkKUq - Bendidi et al. doi.org/10.48550/arXiv.2410.13956 - Wu et al. doi.org/10.48550/arXiv.2408.10609
190
Constantin Ahlmann-Eltze @const-ae.bsky.social · 04/08/2025
I also encourage everyone to checkout the parallel efforts by groups around the world who came to similar conclusions: - @ekernf01.bsky.social et al. doi.org/10.1101/2023.07.28.551039 - Csendes et al. doi.org/10.1186/s12864-025-11600-2 - @kasia.codes et al. doi.org/10.1186/s13059-025-03574-x
1130
Constantin Ahlmann-Eltze @const-ae.bsky.social · 04/08/2025
*We benchmarked scGPT, scFoundation, GEARS, CPA (which claim predictive ability), and scBERT, Geneformer, and UCE (which do not claim this ability). I can't comment on methods released in the last 5 months. For a summary of recent developments, see @ekernf01.bsky.social's post
ekernf01.github.io
A recap of virtual cell releases circa June 2025
In October 2024, I twote that “something is deeply wrong” with what we now call virtual cell models. A lot has happened since then: modelers are advancing new architectures and mining new sources of i...
190
Constantin Ahlmann-Eltze @const-ae.bsky.social · 04/08/2025
Our paper benchmarking foundation models for perturbation effect prediction is finally published 🎉🥳🎉 www.nature.com/articles/s41... We show that none of the available* models outperform simple linear baselines. Since the original preprint, we added more methods, metrics, and prettier figures! 🧵
Beeswarm plot of the prediction error across different methods of double perturbations showing that all methods (scGPT, scFoundation, UCE, scBERT, Geneformer, GEARS, and CPA) perform worse than the additive baseline.Line plot of the true positive rate against the false discovery proportion showing that none of the methods is better at finding non additive interactions than simply predicting no change.
212557
Reposted by Constantin Ahlmann-Eltze
Eric Kernfeld @ekernf01.bsky.social · 28/07/2025
In October 2024, I twote that "something is deeply wrong" with what we now call virtual cell models. A lot has happened since then. How am I updating? New blog post: ekernf01.github.io/virtual-cell...
ekernf01.github.io
A recap of virtual cell releases circa June 2025
In October 2024, I twote that “something is deeply wrong” with what we now call virtual cell models. A lot has happened since then: modelers are advancing new architectures and mining new sources of i...
1122
Constantin Ahlmann-Eltze @const-ae.bsky.social · 03/07/2025
That's amazing. Congratulations 🎉
100
Reposted by Constantin Ahlmann-Eltze
Helena Lucia Crowell @helucro.bsky.social · 27/06/2025
Going from methods dev to full on cancer bio has been tough. All the more excited to see this out at last— www.biorxiv.org/content/10.1... @pascual-reguant.bsky.social @brukerspatial.bsky.social @hoheyn.bsky.social
biorxiv.org
Tracing colorectal malignancy transformation from cell to tissue scale
The transformation of normal intestinal epithelium into colorectal cancer (CRC) involves coordinated changes across molecular, cellular, and architectural scales; yet, how these layers integrate remai...
1125
Reposted by Constantin Ahlmann-Eltze
We Are RLadies+ @weare.rladies.org · 08/05/2025
With {tidylog}, you see friendly messages like: summarise: now 3 rows and 3 columns, ungrouped filter: removed 2 rows (67%), one row remaining ✨ It’s like having a gentle co-pilot, helping you track what each step is doing — super useful for debugging and learning!
With {tidylog}, you see friendly messages like:
summarise: now 3 rows and 3 columns, ungrouped
filter: removed 2 rows (67%), one row remaining
2238
Reposted by Constantin Ahlmann-Eltze
Matthias Meyer-Bender @matthiasmeybe.bsky.social · 05/05/2025
New preprint out! We introduce 𝐬𝐩𝐚𝐭𝐢𝐚𝐥𝐩𝐫𝐨𝐭𝐞𝐨𝐦𝐢𝐜𝐬, a Python package for end-to-end processing and analysis of highly multiplexed immunofluorescence imaging data. Built on xarray and dask, with seamless integration into the scverse ecosystem. www.biorxiv.org/content/10.1...
Spatialproteomics orchestrates workflows to analyze highly multiplexed images. It segments cells, processes images, quantifies proteins, predicts cell types, and provides neighborhood analysis methods, all while integrating into the scverse ecosystem.
1121