the 2024 edition of the Decoding Bio report is live! we analyzed many of the most impactful recent papers in AI, biosecurity, cell-based medicines, immunology, and biotech software. from the jump, a core goal of Decoding Bio has been to clearly communicate why these

if you can compress biology, you can predict biology one way to think about the power of LLMs is as extremely efficient data compression algorithms. at their core, LLMs learn to map high-dimensional input data (e.g., a biological sequence such as a protein) to lower-dimensional

NVentures backed Terray Therapeutics, has unveiled COATI-LDM—a groundbreaking #genAI model for designing small molecules with precision. This innovative example of inverse design has potential to greatly accelerate #drugdiscovery. Check out the arXiv preprint nvda.ws/3MhwqDF

Recursion's clinical trial readouts are not a referendum on the value of AI in drug discovery. some thoughts...

it's our second birthday at decoding bio. help us improve! we're conducting a short survey of our community, only takes a few minutes to complete: docs.google.com/forms/d/e/1FAI…

awesome opportunity for scientist-entrepreneurs that want to build a neuro startup. join us!

🫡 congrats Zavain Dar Nan Li Adam Goulburn

love this framing of venture feedback loops as an RL problem. vc is fundamentally a “wicked” learning environment - the connection between cause (the decision to invest) and effect (investment returns) is complex, feedback is delayed or misleading, and patterns that worked in the

2024 was a huge year for biotech megarounds, with almost 100 startups raising $100M+ financings. shifting so heavily toward these massive deals—and away from smaller, disciplined rounds—is not a good thing for biotech. there are important downsides and knock-on effects for an

agreed. and the best way to improve the biotech talent bottleneck is to give promising, emerging management teams the opportunity to prove themselves vs concentrating most of the capital behind a relatively small number of established executives.

What makes a partnership between pharma and computational drug discovery firms actually work? This session brings together voices from Novartis, Inceptive, Constructive Bio, Terray, and KdT Ventures to unpack how these alliances are built—and what often breaks them. Cynthia

if, for almost any real-world problem, a rational person would rather consult AI than any single human, then AI has crossed the practical threshold we meant when we coined the term AGI.

one frustrating aspect of drug development is that the incentives favor “me-too” programs that chase already-validated targets (e.g., TYK2, GLP-1, PD-(L)1). this is in large part due to the fact that patents protect molecules, not the biology behind them. once a target shows

Are we headed toward a technology revolution in therapeutic BCI? I just wrote this piece about why I believe neurotech will become an indispensable tool for addressing the growing mental health crisis and become as common as personal computing. 🧵 open.substack.com/pub/brainjacob… #BCI

a core tension in applying "virtual cell" models to therapeutic development is the mismatch between the scale at which we measure biology and the scale at which we intervene. we often measure at the cell level (eg single-cell RNA-seq), but we treat at the tissue or organ level

one thing some investors do that drives me insane: ghosting founders. everyone is busy—but a transparent "not a fit" takes seconds. ghosting creates uncertainty, makes managing a fundraising process difficult, and is frankly just a shitty way to treat people. let's do better.