Valentine Svensson
nxn.se
Valentine Svensson
@nxn.se
Molecular cell biology and stats / ML, mostly scRNA-seq.
Principal Scientist at Tahoe Tx.
https://www.nxn.se/archive
Training SCVI - Metal acceleration - www.nxn.se/p/training-s...
January 15, 2026 at 8:12 AM
Reposted by Valentine Svensson
Pertpy: a one-stop and performant framework for single-cell perturbation analysis. @lukasheumos.bsky.social @fabiantheis.bsky.social

www.nature.com/articles/s41...
January 14, 2026 at 4:20 PM
Reposted by Valentine Svensson
PCA is one of those techniques that sounds unreasonable (yeah sure we all want to fit a p-dimensional ellipsoid to our data) but turns out to be a reasonable solution to a wide range of problems.

Also: I do hope stackexchange has been archived, because it is my gen's library of alexandria
January 13, 2026 at 4:13 PM
Citizen Kane is a good movie in general, but the one thing I can't stop thinking about is how thick and substantial the jigsaw puzzle pieces in it are. I've only experienced thin cardboard jigsaw puzzles; these wood ones must feel very satisfying.
January 10, 2026 at 7:54 PM
SCVI - Estimating null expression levels -- www.nxn.se/p/scvi-estim...
January 6, 2026 at 7:57 AM
Finally watched Contagion. Amazing movie! Especially in retrospect.
January 5, 2026 at 4:47 AM
Hoping the mechnical keyboard community at some points discover the 'matter over thread' smart home world. All the current smart buttons are mushy and unsatisfying.
January 4, 2026 at 2:50 AM
Reposted by Valentine Svensson
Why does anyone use the Wilcoxon test for anything?
It's not even transitive. It's sensitive to things one most likely is not looking for (changes in distribution shape) and relatively insensitive to what one usually is looking for (change in location).

www.sciencedirect.com/science/arti...
December 10, 2025 at 9:36 AM
Virtual Cell Perturbation Metrics Reloaded - giovannipalla.substack.com/p/virtual-ce...
Virtual Cell Perturbation Metrics Reloaded
Thoughts on metrics for perturbation prediction models
giovannipalla.substack.com
December 10, 2025 at 4:13 AM
Interpretable Inflammation Landscape of Circulating Immune cells - www.biorxiv.org/content/10.1...
www.biorxiv.org
December 9, 2025 at 4:43 AM
Claude agent skill to infer experimental designs from .h5ad’s - www.nxn.se/p/claude-age...
Claude agent skill to infer experimental designs from .h5ad’s
Anthropic recently launched Claude for Life Sciences, which in practical terms means a marketplace of plugins, MCP’s, and agent skills focused on tasks you encounter in biomedical research.
www.nxn.se
November 14, 2025 at 5:59 AM
What will it take for AI to change drug discovery? - www.writingruxandrabio.com/p/what-will-...
What will it take for AI to change drug discovery?
Some thoughts on avoiding self delusion.
www.writingruxandrabio.com
November 13, 2025 at 7:11 AM
We still can’t predict much of anything in biology - blog.genesmindsmachines.com/p/we-still-c...
We still can’t predict much of anything in biology
Biology is hard. Yes, even for AI.
blog.genesmindsmachines.com
November 12, 2025 at 9:10 PM
The Truth Is Out There, Part 4: The Downside of Belief - www.filfre.net/2024/10/the-...
» The Truth Is Out There, Part 4: The Downside of Belief The Digital Antiquarian
www.filfre.net
November 12, 2025 at 5:35 PM
Hollywood Archaeology: The Super Mario Bros. Movie - grantland.com/features/the...
» Hollywood Archaeology: The Super Mario Bros. Movie
grantland.com
November 11, 2025 at 5:18 AM
Reposted by Valentine Svensson
🧬 Tahoe Workshop at scverse conference 2025 🧬

Disease Biology and Therapeutics in the Age of Frontier Datasets

@tahoetherapeutics.bsky.social

#scverse2025 #ComputationalBiology #DrugDiscovery #TahoeTherapeutics #SingleCell #PerturbationBiology
October 15, 2025 at 2:28 PM
Reposted by Valentine Svensson
MrVI, based on deep generative modeling, is a unified framework for integrative, exploratory, and comparative analyses of large-scale (multi-sample) single-cell RNA-seq datasets. @justjhong.bsky.social @canergen.bsky.social @elhamazizi.bsky.social

www.nature.com/articles/s41...
Deep generative modeling of sample-level heterogeneity in single-cell genomics - Nature Methods
MrVI, based on deep generative modelling, is a unified framework for integrative, exploratory and comparative analyses of large-scale (multi-sample) single-cell RNA-seq datasets.
www.nature.com
October 14, 2025 at 4:44 PM
Training SCVI — Differential expression over epochs - www.nxn.se/p/training-s...
Training SCVI — Differential expression over epochs
With a trained SCVI model you can perform differential expression analysis to learn which genes are increased and decreased in gene expression as cells change their states.
www.nxn.se
October 14, 2025 at 6:14 AM
Causal temperatures - www.nxn.se/p/causal-tem...
Causal temperatures
Time is special.
www.nxn.se
October 5, 2025 at 8:31 AM
Negative binomial regression and inference using a pre-trained transformer - www.nxn.se/p/negative-b...
Negative binomial regression and inference using a pre-trained transformer
I wanted to speed up negative binomial regression with a novel transformer model, but it actually turns out that a classical method of moments is a better solution.
www.nxn.se
September 23, 2025 at 5:54 AM
Hanseatic League - Wikipedia
en.wikipedia.org
September 18, 2025 at 5:06 AM