iiinova — where your digital self keeps evolving

Every single person working today in every branch of cell and gene therapy - viral, nonviral, editing, transgenesis - owes a deep debt of gratitude to Dr. High.

293

I see, thank you. I suspect there will be an arms race in other fields between the watermark placers and the watermark removers....

110

I see. To be clear - I am a complete dilettante in AI - if you use established AI LLMs for image generation - the LLM will embed a non-deface-able watermark?

100

Wow. It should to better. There are 1,536 online textbooks for that it can tokenize and vector-weight-compute on that.

000

The delicate issue is that no one knows how humans know, and there is a growing movement in neuroscience to say the mind learns by tokenization and matrix embeddings. What was the material, K? I tried on genetics and it was flawless.

120

Mmm! In this case perhaps the Licor file?

000

Pedro, journals could ask for the primary Licor files as a countermeasure. WDYT?

100

Fyodor Urnov @urnov.bsky.social · 16/11/2025

Holy mother of my personal lord Gregor Mendel. We now have WHAT?!?

041

Not lower-div. The LLM is superb where there's a huge corpus of outstanding literature about well-understood subjects - AND the user knows how to write intelligent prompts. Try it - go to ChatGPT5 and tell it to explain eg linkage at an upper-div college level. Then ask followups. It's _stunning_.

020

Fyodor Urnov @urnov.bsky.social · 15/11/2025

Re ChatGPT in college ed. I first GSIed in 1991 at Brown and joined Berkeley faculty in 2000 - 50 semesters of teaching upper-div. The "conversations" I've had w the LLM about areas I teach in blow me away. Fellow faculty - specifically for upper-div bio we may be obsolete soon. Don't at me. Try it.

130

Fyodor Urnov @urnov.bsky.social · 03/11/2025

I need to understand how Marco managed to get a camera in front of my car. (those very few of you who do not understand what this is about - enjoy every second of that unique state)

Fyodor Urnov @urnov.bsky.social · 01/11/2025

It's Saturday. With that. I am forced to watch a 35 min UC video about cybersecurity and it's a stunning phenocopy of what, IMHO, is the single best TV comedy work from the UK (other than the Python TV series and FT). Give it a shot. Its joys unfold slowly. www.youtube.com/watch?v=FBaV...

youtube.com

Look Around You: Season 1 Pilot - Calcium

YouTube video by DocShadow660

021

Fyodor Urnov @urnov.bsky.social · 21/10/2025

Please consider La Châtaigne in Lafayette. I came back from two weeks living in the Marais this summer and having the real thing daily. The croissant beurre d’Isigny at La C is as close to the original as one can imagine. Get there early - they sell out fast. They open at 8. 🇫🇷

100

Fyodor Urnov @urnov.bsky.social · 09/09/2025

How incredibly gracious and sweet they named the vehicle "Doc Ricketts." "Canary Row" is one of my favorite books of all time and this story warmed my heart. Plus the fish is totes adorbs. www.nytimes.com/2025/09/08/s...

nytimes.com

A New, ‘Adorable’ Deep-Sea Fish Swims Into View

Fyodor Urnov @urnov.bsky.social · 07/09/2025

Of David's countless accomplishments is the important work that Matt Porteus did in his lab on early-stage genome editing with engineered nucleases (pubmed 12730593). He and Matt both speak about this in the documentary "Human Nature." A true titan. www.nytimes.com/2025/09/07/s...

nytimes.com

David Baltimore, Nobel-Winning Molecular Biologist, Dies at 87

1173

Fyodor Urnov @urnov.bsky.social · 07/09/2025

I will switch out of "dripping with scathing sarcasm" mode and cite a lovely paper by David, Shengdar, Mitch, et al - one of many such papers - where this technically trivial experiment is done. www.nature.com/articles/s41...

nature.com

Potent and uniform fetal hemoglobin induction via base editing - Nature Genetics

A comparison of fetal hemoglobin gene editing strategies using human sickle cell disease donor cells and in vivo transplantation finds that adenine base editing of the –175A>G site in the γ-globin ...

040

Fyodor Urnov @urnov.bsky.social · 07/09/2025

If only, if ONLY there was some method to change these newly discovered TF binding "motifs" at endogenous gene regulatory regions to see if ANY of this new phenomenology has ANYTHING to do with actual gene regulation... 1. PHE PHE SER. 2. SER MET HIS. www.nature.com/articles/s41...

nature.com

Multiple overlapping binding sites determine transcription factor occupancy - Nature

A new method enables comprehensive screening and identification of low-affinity DNA binding sites for transcription factors, and reveals that nucleotides flanking high-affinity binding sites create ov...

230

Fyodor Urnov @urnov.bsky.social · 02/09/2025

"Epigenetics: the way genes learn from experience" (Mikula BC, Genetics 140-379; 1985). Here it is, folks - an atlas of human genes learning from experience. Holy Waddington. www.researchsquare.com/article/rs-7...

researchsquare.com

DNA Methylation Ageing Atlas Across 17 Human Tissues

Aging involves widespread epigenetic remodeling across tissues, yet the nature and consistency of these changes remain unclear. We conducted a meta-analysis of more than 15,000 human methylomes spanni...

091

Reposted by Fyodor Urnov

Nature Biotechnology @natbiotech.nature.com · 26/08/2025

In Brief: A new center in San Francisco will offer tailor-made CRISPR therapies to cure children with rare diseases www.nature.com/articles/s41...

nature.com

Children with rare genetic diseases get CRISPR Cures center - Nature Biotechnology

Nature Biotechnology - Children with rare genetic diseases get CRISPR Cures center

12014

Fyodor Urnov @urnov.bsky.social · 22/08/2025

Fellow editors - nota bene 10x!

030

Fyodor Urnov @urnov.bsky.social · 22/08/2025

PSA: if you're not a mouse person you may enjoy knowing that mouse Sca1 (the hematopoiesis marker) is not encoded by Sca1 (spinocerebellar ataxia 1) but rather Ly6a. I mean, OF COURSE it is. What else could it be? I'm an hg38 person and below is a self-portrait while I was figuring this out.

271

Fyodor Urnov @urnov.bsky.social · 06/08/2025

Around back then I became an ace at dispensing with it 500 ul PB into 20 qiagen columns without splashing chaotrope all over the bench.

030

Fyodor Urnov @urnov.bsky.social · 05/08/2025

+1 My students have a digital one and my reaction is - why?

130

Reposted by Fyodor Urnov

Ryan Cross @scienceboss.bsky.social · 09/07/2025

A new CRISPR Cures center, fuded by $20M from @chanzuckerberg.bsky.social and led by @urnov.bsky.social, will create 8 custom therapies for patients in the next 3 years. “Our goal is to aggressively walk the walk of CRISPR platformization,” Urnov told me. @endpts.com endpoints.news/crispr-cures...

endpoints.news

'CRISPR Cures' center launches with $20M to make customized gene editing therapies

Scientists seeking to make personalized gene editing therapies a reality for more people have launched a new initiative to develop cures for the rarest of diseases left behind by drug companies.

0124

Fyodor Urnov @urnov.bsky.social · 17/06/2025

"Quaternary kissing loops"👌

020

Reposted by Fyodor Urnov

Innovative Genomics Institute @innovativegenomics.bsky.social · 12/06/2025

Thanks to the BBC's The Inquiry program for great conversations with IGI's @jenniferdoudna.bsky.social @doudna-lab.bsky.social and @urnov.bsky.social about Baby KJ, #CRISPR on-demand, and the future of personalized medicine! Listen here: www.bbc.com/audio/play/w... Photo credit: TODAY

083

Fyodor Urnov @urnov.bsky.social · 10/06/2025

Paul Wade was my patient and gracious mentor in Alan Wolffe’s lab (‘97-99). He did gorgeous work back then and continues to do so.

020

Fyodor Urnov @urnov.bsky.social · 10/06/2025

This is v helpful. I'd love to know how many genes are like Car2. Five percent? Ten percent? Because looping is everywhere, and so-called TADs as well, but if 5% of the genes actually depend on looping for their control - well, this means the vast majority of looping is gene-control irrelevant.

100

I missed this paper - this is an important finding - will read. Thank you!

At Vijay's suggestion I looked at your paper (Hansen et al). So would you say Cas2 is the poster child for cohesin-mediated loop essentiality in gene control? B/c clearly for Sox2 it's not true. What other genes like Car2 are this well-characterized wrt this dependency?

110

Stefan, I agree with you completely. Is there evidence that eliminating cohesin-mediated looping affects HBA expression in adult human erythropoiesis? (apologies if I missed it) I am fine with local looping that's cohesin-independent as instructive to transcription.

110

I mean, you kill looping and transcription goes - whatevs. This is a major epistemological challenge.

200