AI-designed antivenom could offer new hope for snakebite victims | REUTERS YouTube video by Reuters

🧪🐍 AI vs venom!
Our Nature paper with David Baker shows AI-designed proteins can block deadly snake toxins. Now featured by Reuters! 🎥
✅ 80–100% survival in mice
💸 Cheap, fast, scalable
🌍 A step toward better antivenoms

📄 www.nature.com/articles/s41...
🎬 youtu.be/gnvEDMEr2mc

We believe this approach can accelerate the development of safer, more effective immunotherapies for patients.
And a special thank you to the talented @mfernandezquintero.bsky.social & Johannes Loeffler for the stunning graphic! 🎨
#CancerResearch #Immunotherapy #AI

This was a monumental team effort bringing together immunology, computational biology, and structural biology.

A huge congratulations to the incredible teams at DTU HealthTech, DTU Bioengineering, & @scripps.edu and to all my brilliant co-authors, especially @Kristoffer Haurum Johansen.

Our platform's key achievements include:
⚡️ Speed: A rapid 4-6 week pipeline from computer design to validated lab result.

🎯 Precision: An AI-driven process to screen for specificity & reduce off-target risks.

🤝 Personalization: It works even for a patient's unique tumor mutation.

In this work, we designed completely novel proteins, or ‘minibinders,’ from scratch. 🤖

These act as a high-precision guidance system, leading a patient's own T-cells (black) to recognize and destroy cancer cells (red) with remarkable accuracy.

#ProteinDesign #GenerativeAI

Thrilled to announce our paper, "De novo-designed pMHC binders facilitate T cell-mediated cytotoxicity toward cancer cells," is officially out in @science.org!

We used generative AI to build a 'GPS' for immune cells to hunt cancer.

Read it here: doi.org/10.1126/scie...

While I’m bouncing between emails and desperately waiting for my holiday, it’s good to know at least something in the lab is holding it together.

Thermal stability might not be glamorous, but it’s underrated.

Designerbodies: even when things get heated, still doing their job ;)

Built for heat. Unlike me;)

Designing binders instead of screening for months? Can’t wait to disappoint reviewers in new and innovative ways.

At AffinityAI we’re moving from “finders keepers” to “designers binders”, one generative model at a time.

If you could have a binder to anything, what would it be?

(Proteins, obscure signalling pathways, that one reviewer who never gets back to you…;)

Ægget redefined the lounge.
We're redefining what it means to design a protein.

At AffinityAI, we combine Danish design sensibilities with precision molecular engineering. Elegant. Efficient. Engineered to bind.

🧬 Discover how we’re reshaping molecular function:

www.affinityai.bio

💡 Curious about:
– How generative AI is transforming protein research?
– Why mass spec is such a rich (but messy) playground for AI?
– What it takes to go from spectrum to sequence to structure?

Give it a listen 🎧

InstaNovo enables diffusion-powered de novo peptide sequencing in large-scale proteomics experiments - Nature Machine Intelligence InstaNovo, a transformer-based model, and InstaNovo+, a multinomial diffusion model, enhance de novo peptide sequencing, enabling discovery of novel peptides, improved therapeutics sequencing coverage...

We focus on our new Nature Machine Intelligence paper introducing InstaNovo — a transformer model for de novo peptide sequencing from mass spectrometry. No reference database needed.
📄 www.nature.com/articles/s42...

AI giver proteinforskningen turboboost AI Denmark · Episode

🎧 Had the pleasure of joining Anders Høeg Nissen on the AI Denmark Podcast to talk about how we’re using AI to crack the protein code 🧬

Catch my segment from 9:30 (and yes — it’s in English 😉)
🎙️ open.spotify.com/episode/4g3R...

Great to see #AffinityAI represented by Esperanza and Oliver at the DTU Startup Day 💪

Lots of interest in what we’re building, especially the fact that we deliver #real #reagents rather than just #AI predictions.

Excited for what's to come!

🚨 We’re hiring a PhD student! 🚨

Join us at DTU Bioengineering + collaborators Messoud & Håkan Ashina + Esperanza Rivera de Torre to bring preventative migraine therapeutics from 💻 to 🧠

🔗 Apply: lnkd.in/dCRU_EN3

🔍 Project: lnkd.in/dbpi8MwX

🚨 Big news! We’ve received a NovoNordisk Foundation Tandem Grant for #MIGRAINE — designing #AI-powered #GPCR-targeting biologics for chronic migraine. From computer to human in 4 years. Proud to team up with Messoud and Hakan Ashina, as well as Esperanza Rivera de Torre!
🧠 + 💊 + 🧬 = new hope.

We’re already supporting:
– Proteins that are hard to purify
– Assays that demand specificity
– Targets that need quantitative detection

Please reach out if you want to hear more.
#StartupLaunch #Founder #AffinityAI

What we offer:
– Designerbodies™ with pM–nM affinity
– ARCANE™ designs for affinity, stability, and expression
– RAVEN™ screens candidates at scale
Built for diagnostics, therapeutics, and biotech pipelines.
#AIinBiotech #SyntheticBiology

🚀 Big news: AffinityAI is live
We’ve built a platform for custom protein binders, AI-designed, in vitro validated, and entirely animal-free.
I'm proud to lead this as CEO.
🔗 affinityai.org

#AffinityAI #Biotech #ProteinEngineering

3/
v1.1 generalises well:
In a GluC-treated HeLa dataset, we reached 81.3% peptide recall—higher depth, same model.

Improved confidence metrics = better precision, fewer false positives, more IDs.
Code: bit.ly/43ExRGI

InstaNovo enables diffusion-powered de novo peptide sequencing in large-scale proteomics experiments - Nature Machine Intelligence InstaNovo, a transformer-based model, and InstaNovo+, a multinomial diffusion model, enhance de novo peptide sequencing, enabling discovery of novel peptides, improved therapeutics sequencing coverage...

2/
What’s new in InstaNovo v1.1?

🧬 +13.5% recall
🎯 +42.6% more exact PSMs
🔍 145% more peptides & 35% more proteins @ 5% FDR
🧪 Support for 7 PTMs: phosphorylation, deamidation, carbamylation, ammonia loss, oxidation, acetylation, carbamidomethylation

Paper: bit.ly/3XBeJFh

Introducing the next generation of InstaNovo models Since our InstaNovo paper is now published, we’d like to share an update on what we’ve been working on while our manuscript was under review. With the release of our preprint over a year ago, we were ...

1/
🚀 InstaNovo just got a major upgrade.

Our Nature Machine Intelligence paper with @instadeepai presents v0.1—but while the paper was under review, we kept building.

Now releasing InstaNovo v1.1, with major gains.
Blog: bit.ly/4lircYB

🧵

I am very excited to announce the publication of this new research article on "Orally delivered toxin–binding protein protects against diarrhoea in a murine cholera model", which is a topic I sincerely care about: nature.com/articles/s41...
Thanks to all co-authors and funders!

Excited to share a new paper in Nature Machine Intelligence that I had the pleasure of contributing to, alongside great collaborators from Europe!

InstaNovo is a deep learning model for de novo peptide sequencing — pushing proteomics beyond databases.

📄 www.nature.com/articles/s42...

Thanks for sharing our work!🎉

InstaNovo enables diffusion-powered de novo peptide sequencing in large-scale proteomics experiments - Nature Machine Intelligence InstaNovo, a transformer-based model, and InstaNovo+, a multinomial diffusion model, enhance de novo peptide sequencing, enabling discovery of novel peptides, improved therapeutics sequencing coverage...

Eloff et al.’s InstaNovo uses transformers plus a diffusion refinement to perform high-throughput de novo peptide sequencing. It recovers more novel peptides at low FDR, outperforms current tools, and extends proteomics beyond database limits. www.nature.com/articles/s42...

Thanks for sharing our work!🎉

AI is helping scientists decode previously inscrutable proteins A new set of artificial intelligence models could make protein sequencing even more powerful for better understanding cell biology and diseases.

New AI tools InstaNovo and InstaNovo+ can now detect proteins previously beyond reach, opening doors in cancer research and disease biology. Early results are promising, though 5% may be false positives. 🧬💻

Thanks for sharing our work!🎉