Bio red-teaming group called mCherry

I interviewed @geochurch and learnt we're already gene editing humans. You just haven't noticed. 0:00 - Gene Editing Mammals → Humans 8:36 - Germline vs Somatic 14:56 - Modified Humans Are Already Here 18:50 - Enhancing Healthy Humans 25:00 - Aging Therapies vs Cognitive Enhancement 30:20 - Embryo Selection 38:10 - Is US Losing To UAE? 42:33 - Biotech Failures 49:31 - Next Dire Wolf Moment 54:21 - AI x Science 1:02:07 - Synthetizing Entire Genomes

Does anyone else start feeling sick after spending too much time with the in vivo data? I went into computation because in vivo breaks my heart. Here I am again 💔

Less model wrappers, more cloud labs.

I drink alcohol sometimes, but only for the dropout regularization

Wow! This is so cool and way more fun than boring BioRender! Plus, developing fine motor skills through hand drawing and creativity without limitations in shapes or form! Try it out yourself here chat.figurelabs.ai/invitatio…

Really excited for 2026. Some trends in Bio x AI that I’m optimistic about are integrated AI design testing platforms and cloud labs. Timelines already shorten when you don’t need to ship materials around or deal with customs.

Maybe my intuition is wrong, but being obsessed with business strategy instead of just building whatever you can build is a huge pitfall.

This week I learned the hard way what a boot manager is 😅

ML/AI in biopharma is anyone’s game. There is no data moat.

NYC is just a series of cubicles and nothing you say can convince me otherwise

🚨🧬 Want to build in drug discovery? Join the M-Boltz Hackathon (Oct 20–21, 2025) with @merckgroup & the awesome Boltz team! Tackle challenges in protein, nucleic acid & drug co-folding, scale cutting-edge models, and build the next wave of open science. ( get to hang with @GabriCorso) 🌍 Hubs in Darmstadt & Boston remote 💡 Sponsored by @huggingface, @nvidia , and @awscloud 🔗 Register now: moml.mit.edu/m-boltz-hackath…

So many flow/pipelining software packages, so little time...

Bioinformatics lore of the day: the nf-core metro maps are based off of the Tube. 🚇

There is no replacement for just trying/doing things, especially in biology. So many experiments have unexpected outcomes. Stop trying to estimate what “should” happen and observe reality. It is always more interesting and complex.

Only real biologists own this pen.

SuperGrok is best for bioinformatics, followed by Claude 3.7. I think this is probably because SuperGrok actually pulls the documentation and gives you more or less the proper syntax. All others give overly complicated solutions or breaks pipeline syntax often.

One day we’ll look back at animal studies in disbelief over how barbaric we were.

send me a message, i take a nap call that a rest api