Tanmayee Narendra

This was never really the case. We have known since the 1960s that some of our non-coding DNA has a crucial role in gene regulation: turning the expression of coding genes on & off. It’s been clear for decades that the regulatory DNA is at least comparable in proportion to the protein-coding DNA. /5

Great thread!

So there you have it: AlphaGenome is a great start, and will surely be a valuable tool. Whether it will lead to clinical advances remains to be seen. Its applicability will be limited by its very nature. And we still need to do the basic science. 32/32

So, what has AlphaGenome done? As with AlphaFold and the protein-folding problem, it has sidestepped developing a full scientific understanding of the complexities by using AI to infer statistical associations – in this case, between genomic variation and biological outcomes. /19

I cannot endorse this enough. AI itself is not "taking us" anywhere. But its developers would certainly like to.

Now out in NAR Genomics and Bioinformatics!

academic.oup.com/nargab/artic...

Thank you to the organizers for bringing so many diverse viewpoints together!

@barcelonacollaboratorium.com @crg.eu

Causal Discovery from Conditionally Stationary Time Series Causal discovery, i.e., inferring underlying causal relationships from observational data, is highly challenging for AI systems. In a time series modeling context, traditional causal discovery methods...

I was at the Symposium on Causality in Biology & AI in Barcelona. I presented a flash talk on inferring time-varying gene regulatory networks from time series scRNA-Seq data.

Some results here - arxiv.org/abs/2110.06257

This is not a HiC map! Ever wondered if multiple enhancers get activated simultaneously? We measured chromatin accessibility on thousands of molecules by nanopore to create genome-wide co-accessibility maps. Proud of @mathias-boulanger.bsky.social @kasitc.bsky.social Biology in the thread👇

Very cool tool name!

🤝 It was a real joy to reconnect with my PhD supervisor and mentor @gjbarton.bsky.social, as well as former colleagues
@stuartmac44.bsky.social, @foreveremain.bsky.social, @mitsenkov.bsky.social, Vanessza Fentor, Douglas Freeburn,
@tanmayee.bsky.social and @rosiegallagher.bsky.social!

Causal Discovery from Conditionally Stationary Time Series Causal discovery, i.e., inferring underlying causal relationships from observational data, is highly challenging for AI systems. In a time series modeling context, traditional causal discovery methods...

Update: Our paper on "Causal Discovery from Conditionally Stationary Time Series" was presented two weeks ago at ICML 2025!
arxiv.org/abs/2110.06257

One thing that really bothers me with the new "virtual cell" terminology is that it is currently largely focused on a very narrow definition of models that can predict effects of trans perturbations (gene dosage, drugs etc) on gene expression. 1/

We released our preprint on the CREsted package. CREsted allows for complete modeling of cell type-specific enhancer codes from scATAC-seq data. We demonstrate CREsted’s robust functionality in various species and tissues, and in vivo validate our findings: www.biorxiv.org/content/10.1...

Lungfish xkcd.com/3064

🚨New pre-print on Biorxiv! 🚨

Histone modifications are measured with ChIP-Seq, but the chromatin landscape introduces biases. We introduce DecoDen to simultaneously learn shared chromatin landscapes while de-biasing individual measurement tracks.

Check it out: github.com/ntanmayee/de...

🧬 🖥️

Towards Personalized Epigenomics: Learning Shared Chromatin Landscapes and Joint De-Noising of Histone Modification Assays

Figure 1

Figure 2

Figure 3

Towards Personalized Epigenomics: Learning Shared Chromatin Landscapes and Joint De-Noising of Histone Modification Assays [new]
Learns shared chromatin structure, correcting histone modification biases across epigenomes.

Towards Personalized Epigenomics: Learning Shared Chromatin Landscapes and Joint De-Noising of Histone Modification Assays https://www.biorxiv.org/content/10.1101/2025.03.04.641154v1

Towards Personalized Epigenomics: Learning Shared Chromatin Landscapes and Joint De-Noising of Histone Modification Assays https://www.biorxiv.org/content/10.1101/2025.03.04.641154v1

Digital sovereignty – Huber Group @ EMBL The US is now a mafia-style extortionist you want to stay away from

It's time for individuals and organizations in Europe and other democratic countries to move mission- or life-critical IT services and infrastructure outside the reach of the US government; noting that US tech companies have already largely subordinated themselves.

www.huber.embl.de/group/posts/...

Three AI-powered steps to faster, smarter peer review Tired of spending countless hours on peer reviews? An AI-assisted workflow could help.

Let's talk about this Nature piece in more detail.

I've rarely read something so anti-scientific anywhere short of the National Review.

www.nature.com/articles/d41...

How do ancient genes regulate animal embryo development at single cell resolution | University of Dundee, UK A PhD project at the University of Dundee

Please spread the word🙏:

[PhD Position in Computational Evolutionary Transcriptomics]

If you are interested in doing a PhD in gorgeous Scotland on 'Why embryo development goes wrong sometimes?', please consider applying and join our wonderful team in Dundee!
www.dundee.ac.uk/phds/opportu...

RNA xkcd.com/3056

I've been developing a semantic search tool that covers not just bioRxiv and medRxiv, but the entire PubMed database. This means you can search across a massive collection of biomedical research using keywords, questions, hypotheses, or even full abstracts. Try it out: mssearch.xyz

(1/4) Thrilled to announce another major release of the HOCOMOCO motif collection, well-known for its silly name and rigorous approach to constructing and benchmarking DNA sequence motifs recognized by human and mouse transcription factors.

hocomoco.autosome.org

What's Happening Inside the NIH and NSF

A long post about what’s happening to the science funding agencies in the US and why. As mentioned, this one just kept getting longer even as I kept stripping curse words from it.

www.science.org/content/blog...

How should the advancement of large language models affect the practice of science? | PNAS Large language models (LLMs) are being increasingly incorporated into scientific workflows. However, we have yet to fully grasp the implications of...

An LLM "creates textual claims, and then predicts the citations that might be associated with similar text. Obviously, this practice violates all norms of scholarly citation.

At best, LLMs gesticulate toward the shoulders of giants."

Bender, West, and I contributed to this pro/con piece in PNAS.

STEM Collectives: Sisters are doin’ it for themselves - Rukhmabai Initiatives In the uneven playing field of academia and research, women are now empowering women through mentoring collectives that connect experienced people with those seeking guidance and support in education,...

Anagha Pavithran interviewed me and Janani about mentorship programs for Indian women interested STEM research. The article is finally out - rukhmabai.com/stem-collect...

Thank you for your kind words about WISR!

Glad to see other organisations like WoAA also mentioned :)

Hello! Could you please add me to ML for single cell biology?

Great work from our colleagues in the Division of Computational Biology! Gabriele Schweikert's group work on machine learning techniques in epigenomics. scholar.google.com/citations?hl...