We wrote about a topic that we care a lot about at Flory and is probably what's on everybody's mind :)
Tl;dr: AI for drug discovery has meant something narrower than what it should be, we might have been looking at solving the wrong variable, and opportunities in biotech are far from dead but increasingly concentrated in a few wild plays.
Debates welcome.
Kinda obsessed with the new COVIB cryptic pocket paper;
Normal mode analysis is obv not new but this seems good enough as a cheap proxy for finding alternative conformations / cryptic pockets.
Raises a very interesting question of how much resolution we actually need for driving decisions in drug design.
Imo thermodynamic distribution is still critical at the end of the day, but there could potentially be a super powerful use case where we combine COVIB w some medium/high-throughput experimental input like photoaffinity fragment mapping / MS
https://t.co/XxLF9CfRnI
At 23 I turned down a PhD at Cambridge and walked away from a peptide spinout I’d spent the previous year building to start my own company.
I think your risk tolerance will be this fucked up too when you almost got kidnapped as a kid over business 😅
We think there’s a fundamentally different way to design drugs. Don’t start by designing a good binder. Go straight for what will control what the protein does.
We’re going after some of biology’s hardest targets and diseases we still can’t cure.
First post on the wild idea we’re after here. Tell me why we’re wrong :)
https://t.co/ROaPhy7xdf
I think science needs a bit of naivety.
A lot of good ideas start with dumb questions that sound ignorant or obvious. Most go nowhere; but knowing too well how things are “supposed to work” is its own failure mode.