Example of how precarious a position the off-shoring CRO model has left biopharma innovation in.
Silver lining: breaking the incremental labor arb addiction forces an overdue rethink of how synthesis / research broadly gets done just in time to re-orient that foundation for AI.
It's time for all drug developers to reconsider heavily relying on Wuxi to synthesize their proprietary compounds
How $ERAS obtained ERAS-0015 from Wuxi>Joyo>$ERAS should be a case study
$RVMD used Wuxi to synthesize their very early pan-RAS compounds, and guess what? Wuxi imminently came up with the nearly identical structures as Darax and tried to patent that...
Concrete example in small mol chemistry / synthesis GNNs is long-range functional group interactions getting “squashed.”
e.g. a distal electron-withdrawing group or bulky substituent can completely change reactivity or binding at a site several bonds away but standard message passing compresses that signal into a few hops of uniform aggregation.
Transformers or hybrids (even though graphs seem like the right inductive bias) tend to outperform if they preserve those nonlocal dependencies instead of diffusing them.
AI Scientists "discuss science, not do it" @sajithw & @benchling get it.
AI for Science is sandwiched by physical layers both upstream (train = move+measure atoms) & down (deploy = manipulate+make).
Who connects those (and where they do it) matters a lot. @satomic_ai is on it.
In the AI era, the traditional biopharma industry is the underdog. Big tech and AI labs are building wet labs. China has overtaken Europe in molecules produced. But the tools available to the industry discuss science, not do it.
The hard problem in AI for science is at the interface between the physical and digital worlds.
We built an AI Scientist at that seam. It wires together the digital and physical worlds of R&D. Predictive models, data infrastructure, wet lab execution feed into a single loop that reasons, acts, and improves with every experiment. Our ambition: get molecules to the clinic twice as fast.
Last fall I wrote about why biotech needs to be rebuilt for the AI era. Today I'm sharing the next chapter: what the AI Scientist is, a blueprint for how it works, and why even Richard Feynman couldn't hack it in a wet lab.
15/We're talking to design partners, and will be shipping full production orders this fall. Please reach out if you want more molecules, faster, or if you want to join us to work insanely hard against one of the most important challenges of our time.
1/I'm not a scientist, but have spent my career learning that the most important thing you can build in life sciences isn't a drug, but the infrastructure that makes better drugs inevitable.
Today we're starting to talk about what we're building at @satomic_ai to do just that.
14/We've raised a $15 M seed round with fantastic investors to build the world's most capable synthesis platform on-shore, with drug-like chemistries, and faster than anything on the market.