My take-away from #SMBE2026 after 3 days: biodiversity genome projects such as Darwin Tree of Life @ergabiodiv.bsky.social @vertebrategenomes.bsky.social et al are no longer promises, but are changing how we do molecular evolution. I hope biodiversity gene expression will be next 😎 .

Come and hear @davidolusoga.bsky.social and I chatting about history, genetics, race, that sort of thing, to mark the 10th anniversary of my book A Brief History of Everyone Who Ever Lived. unionchapel.org.uk/whats-on/ada...

Really incredible work from the Tcheandjieu Lab that finally helps us to decipher what is going on at the 9p21 locus and its association with coronary disease on different ancestral backgrounds. www.cell.com/ajhg/fulltex...

🗞️Sortim al 1º 'Va de Ciència' del @ara.cat! Gran reportatge dels primers 25 anys des del genoma humà amb l'Ivo Gut (CNAG), @ewanbirney.bsky.social (@embl.org), @rodericguigo.bsky.social (@crg.eu), @tmarquesbonet.bsky.social (@upf.edu) i @gmarfanyn.bsky.social (@ub.edu) www.ara.cat/ciencia-medi...

“Doubly uniparental inheritance and sex determination evolving multiple times in mussels” has entered the chat

One of the oddest things to get across to non-genetics/genomicists is just how weird and varied hetreogametic sex development pathways are. You just ... could not make it up. (eyes sideways to the platypus - in particular you!). So - ribosomal protein as sex determining factor. Just normal crazy.

Most diseases are caused by a combination of factors. Rama Abdulhamid, PhD student in the Stegle Group at EMBL Heidelberg, uses AI to study thousands of such factors at once. By using AI to spot patterns in how a disease develops, Rama and her team can find the health factors driving it. #EMBL_AI

For my German friends and colleagues - that was well played and I think you should have won that game. Goal wrongly dissallowed but you can't fault the passion and nerve of the Paraguayans

Feeling for Germany. Fingers crossed on penalties.

Congratulations to Prof Richard Durbin, winner of the 2026 Genetics Society Medal! His pioneering work in computational genomics has transformed how we analyse and understand genome sequences.

📣 Out now @naturemedicine.bsky.social 🧬🤖 👉 rdcu.be/fqasl Our automated reanalysis tool Talos enables broad adoption and delivers timely + equitable #raredisease #diagnosis at scale! ♻️ Runs monthly on >10K datasets 🎯High specificity 💰Low running costs 🤗Open source @dgmacarthur.bsky.social

This is just one of the best bits of work I’ve been involved with recently and a tour de force in comparative single cell development - exploring when, how and why organisms develop different - in this case two fish hats off to Jess @bournsupremacy.bsky.social, Mike @mwdorr.bsky.social and team !

It is really interesting to chart the increase in sophistication and depth in the rare disease community as we bring forward the "genotype early" or even "genotype first" diagnosis pathways, where as "genotype late" was the more common approach >>>

👏👏👏 fantastic to see @nickywhiffin.bsky.social win the Leena Peltonen award at #eshg2026 @eshg.bsky.social 👏👏👏

A major epigenetics study using UK Biobank samples will unlock new insights into the biological basis for human health & disease ✅ Led by @exeter.ac.uk the research will transform how we understand, predict & treat health outcomes including heart conditions, dementia & cancer.

Congratulations to @gautamdey.bsky.social and @dudinlab.bsky.social - so well deserved

22/n On a personal note, I’ll soon start an independent research group in the Genome Biology Unit at EMBL Heidelberg. If you enjoy this type of work and want to engineer genomes at the largest scale, please reach out! jonas-koeppel.github.io/koeppellab/ www.embl.org/groups/koepp...

📢I'm setting up a new team at the @humantechnopole in Milan. We have several openings in AI applied to nationwide health data, biobank-scale -omics. Milan is a beautiful city and at @humantechnopole is top notch. You can find post-doc positions here: careers.humantechnopole.it/job/Postdoc/...

The fact we can see this machine factory frozen in time (and space), happily making molecular machines, effectively at an atomic level is just jaw dropping. Every time I see these images I am blown away by it. Great work by Xiaohan and what a way for Christoph Müller to retire from @embl.org

'Species' is a useful term, but one that we have imposed to make sense of the diversity of life. Nature deals in genetic lineages, haplotypes and ancestral recombination graphs. Speciation researchers should be (and increasingly are) doing the same. And using as much genomic data as they can get!

Last talk before lunch - Steve Quake - His talk title: How many cell types arise from one genome? - he gives a teaser that he wont give an answer. Then the famous Brenner quote "Biology is drowning in a sea of data and starving for knowledge".

Peter Koo moves us onto virtual cells - for him going beyond transcription. Peter calls the DNA -> mRNA as seq2fun and then cell state in transcription as a separate. Putting everything into one model is complex because the hetreogenous nature of inputs

Jian Zhou (U Chicago) on moving to single molecule readouts (Fiber-seq and friends - the community has not settled on one term yet I think!). Needs a different model - not sequence to levels but sequence to probability distributions. This...works (though it is interesting to think about how to test)

So - now to Stein Aerts on modelling genome regulation. Talks us through a similar framework but moves into synthetic models for fly and zebrafish and beautiful readouts. Uses a VAE (variational autoencoder) representation of the seqeuence motifs, and then extends to enhancers by combinations

Now David Kelly, Calico, of Borzoi fame on his omnibus model. He shows some neat transfer learning results. Zoom into PolyA site predictions in particular Brain non Brain (so called Borzoi prime because it has 3' data, but it is "just another track"). This basically ... works.

Now to Anshul Kundaje @anshulkundaje.bsky.social - (another legend!) - on decoding regulatory DNA. He starts with the classic epigenomics roadmap cube (makes me nostalgic) but it's the same basic problem as AlphaGenome (editorial - Anshul is going to have a very different take on this) #cshlsymp26

Ziga introduces the AlphaGenome - emphaising that one wants to have both basepair resolution (like ChromBPNet) and long range (like Enformer/Borzoi). Talks about clever batching with engineering on the GPU; Encoders on small segments to downsample and then a transformer layers on internconnecting

Made it to #cshlsymp26 and will be doing some intermittent posting on the AI symposium here - starting with the wonderful Žiga Avsec from Google Deep Mind on AlphaGenome (editorial - it is rather great to see many Europeans - Londoners included! - at this meeting)

We're happy to announce our new preprint! 🐸 easymode: general pretrained networks for cellular cryo-ET. Segment ~20 cellular features – ribosomes, microtubules, mitochondria, nuclei & more – with zero model training. 🔗 doi.org/10.64898/202... 🧵👇

1\ Excited to finally share some of my PhD work: OpenSplice! We measured how >590,000 mutations affect exon inclusion across 600 human exons, creating a resource to study splicing regulation, benchmark variant effect predictors, and support variant interpretation.

Gosh. Just letting it all sink in. North London forever...

This is one of the best books introducing genetics, genomics and how this relates to how humans have populated the world. It's fun, scientifically accurate and interesting. (and I feature as someone who "pratted around" in my youth in this world. Sort of true).

🚨 New preprint! How does noncoding variation drive complex traits? We built a massive atlas of cattle regulatory variants (>150k emVars!) and directly tested them in both bovine AND human cells. 🐄🧬🚶‍♂️ Read it here: www.biorxiv.org/content/10.6...

Nextstrain #ANDV trees updated with a few improvements: - 2 outbreaks marked - host scientific/common name added - province now visible on map - date colouring added Here's a quick tour of how to change colours, geography & names! Links to trees here (scroll down): nextstrain.org/groups/hodcr...

For me this is the most common oversight in grant applications which is simple to fix - explain why *you* *the investigator* believe that the study you are proposing is going to work. If someone has done it before at this scale but different samples/disease/context - brilliant! tell me!

First genome sequence from Andes hantavirus strain from ongoing ship outbreak posted on Virological: virological.org/t/complete-s... courtesy Swiss National Reference Centre for Emerging Viral Infections, Geneva University Hospitals, University of Zurich

UCL really does boast some of the best science communicators on the planet.

In collaboration with the TransCODE consortium, EMBL-EBI scientists have found over 1,700 new proteins in the ‘dark proteome,’ a region of the human genome once thought to be noncoding. These new proteins could have implications for human diseases, including cancer. tinyurl.com/yc4t6s3s

Minflux is one of the technologies pushing the boundaries of what we can see in living cells - tracking single molecules at nanometer resolution in living cells. Sounds like science fiction - is actually doable @embl - come and learn more!

(1/15) Thrilled to share new Evo-Devo work where we address the molecular, cellular & morphometric basis of an extreme axis segmentation program in the Japanese eel — an animal that forms 120 vertebrae.

5 mins for Arsenal’s season … made on Merseyside….

Suddenly a toffee fan. Amazing game.

UK Biobank has my data, but I’m not worried. I know the benefits are too great to consider pulling out www.theguardian.com/commentisfre...

The hill has Paleolithic rock shelters occupied by Neanderthals approximately 50 thousand years ago, long before the Iron Age hillfort was built. One of the last known Neanderthal sites in Britain

If you're interested in the history and philosophy of randomization in all its forms I'm co-organising a *FREE* conference in glorious Bristol 2-3rd July wth the legendary Samir Okasha, registration is here www.eventbrite.co.uk/e/randomizat... If you think it looks interesting please advertise

Bluesky uk energy transition hive mind - could today be the first 0 fossil free 30mins?

Grant writing tips from a grumpy reviewer ploughing through grants. 1. Often your reviewers are going to come from a broad range disciplines adjacent but *not actually in* in your field. Help us get orientated in the summary with an analogy/sketch and go back to analogy / sketch throughout.

I wrote a friendly rant about the use of causal language in AI. www.linkedin.com/posts/ewan-b...

First Spargel of 2026 - with hefeweizen of course in beautiful Heidelberg with @oliverstegle.bsky.social - love the German spring - with extra AI love!

Are you ready to play our special version of pinball? 🕹️ This cover of @natrevgenet.nature.com was inspired by our review on the interpretation, extrapolation and perturbation of single cells. The artwork takes the form of a pinball machine, as a deliberate reimagination of Waddington's metaphor.