@developers.google.com Benchmark performance is one thing, but reliable systems that devs can rely on are another. Gemini 3 API has been broken for ~14 hours now (responding in JSON, not natural text), with dozens of reports across Gemini forum, GitHub... Yet no solution. Please escalate internally.

Proud to close out an incredible year with an update on where scverse is today and where we’re heading. I’d love for you to join me, share your ideas, challenge us, and help shape what comes next. It’s time! 💪

AI holds huge promise for life science research. But unreliable answers currently limit its impact. @juliosaezrod.bsky.social, Head of Research at EMBL-EBI, discusses how Model Context Protocol (MCP) servers are helping to change this. www.ebi.ac.uk/about/news/p... 🧬💻

What a suitable moment to have a community meeting by @maltekuehl.com and @biocontext.ai! Since we didn’t manage to have him on the conference, join us next Tuesday! 💪❤️

We will have our next community meeting on Tuesday, 2025-11-25 at 11:00 CST! Malte Kuehl (@maltekuehl.com) will be presenting on BioContextAI! More information and Zoom registration link in 🧵

🤖🧬 New biomedical MCP server added to the registry: Nucleotide Archive MCP server Check it out at https://biocontext.ai/registry/biocontext-ai/nucleotide_archive_mcp

🤖🧬 New biomedical MCP server added to the registry: anndata-mcp Check it out at https://biocontext.ai/registry/biocontext-ai/anndata-mcp

The protein concentration in the cytoplasm is so high that the average protein has a water hydration shell with a thickness of only ≈ 10 water molecules separating it from the adjacent protein hydration shell. 1/2

📄 BioContextAI is now slightly updated and published @natbiotech.nature.com ➡️ www.nature.com/articles/s41...

BioContextAI is a community hub for agentic biomedical systems - @puellesv.bsky.social @biocontext.ai go.nature.com/3JmqKux

Yet another reason to build with Python

Why does the naked mole rat have the longest lifespan of any rodent, nearly 40 years? A 30-year long mystery unraveled @ScienceMagazine today! Its cGAS enzyme in cells has 4 missense mutations that upends its function, promoting DNA repair and suppressing inflammation www.science.org/doi/10.1126/...

Recently, @anthropic.com introduced Claude Skills, including some for the life sciences. But how can you use skills in your own agents that may not support them out of the box? The answer is our new Skill-to-MCP server. More info below ⬇️

People outside tech starting to realize how quality software and typing/generating code fast are not correlated

A truly helpful resource for all things omics! Make sure to check out their MCP server, too: biocontext.ai/registry/sae...

It was a great pleasure to collaborate with this team - a quite different perspective on spatial omics, applying lessons from other scientific fields to biomedicine with robust spatial statistics. Thanks to everybody involved! Tutorial available at: robinsonlabuzh.github.io/pasta/00-hom...

It was a great pleasure to collaborate with this team - a quite different perspective on spatial omics, applying lessons from other scientific fields to biomedicine with robust spatial statistics. Thanks to everybody involved! Tutorial available at: robinsonlabuzh.github.io/pasta/00-hom...

💫 The 2025 SIB Innovative Resource Award goes to @scverse.bsky.social 💫 The jury is impressed by the focus of these omics tools on interoperability, usability, and communication. The project’s solid governance and emphasis on open science principles are particularly commendable #SIBawards #bc2basel

Love the social media efforts and community building of @scverse ! https://scverse.org/join/ https://scverse.org/events/ #bc2basel

Interesting work by @slobentanzer.bsky.social and team, making BioCypher available through MCP! Check it out on biocontext.ai/registry/bio... or try it in BioContextAI Chat (biocontext.ai/chat, click "cogs" icon to add the remote MCP server).

Interesting thoughts on using specialized small models for agentic systems. Potential to cut cost and build more sustainable and privacy-preserving local-first solutions? arxiv.org/html/2506.02...

🤖🧬 New biomedical MCP server added to the registry: Holy-BIO-MCP's gget server Check it out at https://biocontext.ai/registry/longevity-genie/gget-mcp

Nature research paper: Pathology-oriented multiplexing enables integrative disease mapping go.nature.com/3HdZxZG

Glad to contribute to this collaborative community platform. A case study shows MCP servers working together - combining BioContextAI Knowledgebase with our omnipath omnipathdb.org MCP (work in progress) to showcase interoperability. Looking forward to see how the ecosystem evolves!

Preprint alert 🚨 Do you use chatbots in your work or even build MCP servers and agentic systems yourself? Or would you like to find a way to use biomedical tools using natural language? Then check out biocontext.ai, now out on bioRxiv: www.biorxiv.org/content/10.1...

Preprint alert 🚨 Do you use chatbots in your work or even build MCP servers and agentic systems yourself? Or would you like to find a way to use biomedical tools using natural language? Then check out biocontext.ai, now out on bioRxiv: www.biorxiv.org/content/10.1...

Community-based biomedical context to unlock agentic systems https://www.biorxiv.org/content/10.1101/2025.07.21.665729v1

🚨Scaling multiplexed imaging 📈 We are excited to share Pathology-oriented multiPlexing (PathoPlex). Now out in @nature.com: www.nature.com/articles/s41... 🧵Walk-through thread below ⬇️

🎉 The revised version of CORNETO, our unified Python framework for knowledge-driven network inference from omics data, is published in peer reviewed form 🔗 Paper: www.nature.com/articles/s42... 📖 News & Views: www.nature.com/articles/s42... 💻 Code: corneto.org 🧵 Thread 👇

🎉 Such a great work by everyone involved in this major push forward in spatial multiplexing and next-generation pathology. I‘m glad to have been able to contribute to this effort and shed a light on the discovery of sub-cellular to tissue level organization patterns by xAI based on this technology.

Hamburg News: ⁦@PuellesVictor⁩ and collaborators push tissue analytics to the next frontier. This also marks a remarkable academic path: from postdoc to junior professor in our group to full professor at ⁦@UKEHamburg⁩ & ⁦@AarhusUni⁩ ⁦‪@Nature‬⁩ www.nature.com/articles/s41...

A new imaging technique, PathoPlex, enables simultaneous analysis of over 100 proteins in tissue samples, offering detailed mapping of disease processes and potential for broader clinical applications. doi.org/g9tnd4

Huge shoutout to @maltekuehl.com &Co for this new publication! Amazing work, exciting science, and well deserved recognition 👏

Pathology-oriented multiplexing enables integrative disease mapping @nature.com www.nature.com/articles/s41...

I can't imagine a better day of the week to publish this than #FluorescenceFriday: PathoPlex is finally published! 🎉 This presents a revolutionary jump from 4-6 to 100+ markers per sample! 🤯 Read the full story here: www.nature.com/articles/s41...

#HotFromThePress !! What if instead of analyzing 4-6 pathology markers you could check >100 in the *same* sample? @maltekuehl.com & co @puellesv.bsky.social lab introduce #PathoPlex, method that could even predict disease Full www.nature.com/articles/s41... @focalplane.bsky.social @nature.com 🧪🔬

🚨Scaling multiplexed imaging 📈 We are excited to share Pathology-oriented multiPlexing (PathoPlex). Now out in @nature.com: www.nature.com/articles/s41... 🧵Walk-through thread below ⬇️

It's here! ✨ The #ExSRRF protocol paper is out 😍. Full details for merging Expansion Microscopy with our eSRRF analysis. From cells to tissue. A partnership with @puellesv.bsky.social's lab. Lead by Dominik Kylies and @hannah-superres.bsky.social + colleagues. #ExM #eSRRF ➡️ Open link: rdcu.be/euvEw

Excited to share our protocols paper on combining expansion microscopy with super-resolution algorithms! #ExM #ExSRRF

No you cannot determine which drugs are safe and effective with machine learning. You cannot. It will not work. People will die.

There is absolutely no reason to think that such models are actually learning biology. So maximizing the perturbation prediction objective does not necessarily result in a "virtual cell". 3/

Update: We greatly revised our paper and renamed it “Harnessing the Potential of Spatial Statistics for Spatial Omics Data with pasta”. We discuss the broad range of exploratory spatial statistics options for spatial Omics technologies and show relevant use cases. arxiv.org/abs/2412.01561

Confounders are a major bottleneck to understanding biology. Interactions confounded by unobserved regulators (variables) cannot be causality inferred. Gazillion data points from the observed variables cannot compensate for unmeasured confounders. What can help ... 1/2

Unsurprising finding, but glad to see someone be so comprehensive! Genomics data is SUPER structured and biased and so you need to be extra thoughtful when doing ML. Zero-shot evaluation reveals limitations of single-cell foundation models genomebiology.biomedcentral.com/articles/10....

Happy to see our overview of nanoscale imaging techniques for biology out!

RNA abundance is regulated by well-established mechanisms that do not alter chromatin accessibility. I am not sure how to reconcile these mechanisms with the claim in the abstract of this article: 1/n

in case you're wondering, "what's the harm in claiming an extinct species has been brought back from the dead" when it most certainly has not, our interior secretary is already using it to justify taking animals off the endangered species list

Not to be pedantic here, either, but "AI" is not taking his readers. Google and OpenAI and Perplexity are taking his readers, specifically to drive up their user bases and ad sales, on the back of his (and countless others') work.

1/ New preprint! 🍳 @elihei.bsky.social and our team at @embl.org , @dkfz.bsky.social, and @mskcancercenter.bsky.social built #segger - a fast, accurate cell segmentation tool for spatial transcriptomics that assigns transcripts to their cell origins! doi.org/10.1101/2025...

Cell2fate learns which genes have co-regulated transcription rates => spatially mapping velocity modules & improved robustness. Bayesian modelling => easier to add biological realism. Exciting collab with Alexander Aivasidis in Bayraktar & @steglelab.bsky.social www.nature.com/articles/s41...