Twitxy Saved

Jacob Glanville

Verified account This account is verified. Learn more

@CurlyJungleJake

Founder & CEO of Centivax
San Francisco, CA centivax.com Joined May 2017
232 Following 27K Followers 4.8K Posts

Latest posts 5

Pinned post

Jacob Glanville Verified account @CurlyJungleJake
In a world of slop and 10 second sound bytes, this interview with @NikoMcCarty is your antidote. If you want to hear what has been by far the best researched and most detailed interview I’ve done to date, check out this video where we go really deep on universal vaccines and universal antivenoms. Now excuse me while I go follow Niko’s interview channels. #universal #vaccines #antivenom #eradication
Niko McCarty. Verified account @NikoMcCarty Oct 29
New episode on universal flu vaccines (and antivenoms) with @CurlyJungleJake, CEO of @centivax. They're trying to create a vaccine that protects against all strains of influenza: past, present, and future.

This is obviously a difficult thing to do. Seasonal flu vaccines are designed to protect against a virus that constantly mutates and recombines in animals. The efficacy of seasonal vaccines is also quite low; ranging from about 15 to 60 percent.

Instead of making vaccines that protect against only a few circulating strains of the virus, Centivax is trying to teach the immune system to recognize parts of the virus that many strains share and that resist mutation. The hope is that antibodies targeting these regions could protect even against strains we haven’t encountered yet, without scientists having to reformulate the vaccine every year.

Things seem to be going well: the company has tested the vaccine in about 180 people so far and will soon begin a Phase 2 trial.

This is probably the most ambitious technical interview I’ve ever done. We cover some extremely deep immunology, but Jake explained the ideas really clearly, and I left feeling like I finally understood how vaccines work and how scientists make them. I also came away much more convinced that a universal flu vaccine could work.

00:49 — Why flu vaccines need constant updating
03:07 — How influenza infects a cell
06:39 — Why our bodies don’t always make broadly-neutralizing antibodies
16:07 — Mutations and animals
23:42 — What is vaccine efficacy?
28:55 — Designing a universal flu vaccine
37:40 — The idea behind Centivax
44:16 — Clinical trials
57:33 — The business case for eradication
1:03:26 — How to eradicate the flu
1:07:26 — Could we make broad vaccines for the common cold?
1:09:07 — Building a universal snake antivenom
1:23:09 — Antivenom cocktails
View quoted post on X
3 replies 4 reposts 29 likes 3.8K views View on X
Jacob Glanville Verified account @CurlyJungleJake
AI is super useful in accelerating things scientists can already do, but the hyperbole of some of the posts are driving actual scientists nuts. This is literally what these posts read like to us: “stunningly, for the first time ever, AI has enabled the recipe for chocolate chip cookies. With this cookie code break complete, today marks a revolution in disrupting culinary industry and unlocking an unimaginable future of dessert abundance.” It’s a disservice to the public to overhype nondisruptive demos.
5 replies 11 reposts 135 likes 5.3K views View on X
Jacob Glanville Verified account @CurlyJungleJake
Read the fine print. This claims a $1M Phase 2 trial. What it actually contains is a 24 person Phase 2a repurposing an already approved drug that only lasted 2 weeks. Critically underpowered. The submissions are a mess, with a huge delay of registration over 100 days after study start, parallel vs crossover discrepancies, an unfavorable finding from ethics board, and notably a failure to report the results in January as was committed. They benchmark against trials much larger than theirs to make their numbers seem lower in a misleading way. The study uses an existing oral medicine, one virtual site, home-based monitoring, and a very small patient population. These are all important cost-reduction strategies, but none depends uniquely on generative AI. I’m supportive of AI reducing trial costs but this is wildly misleading and unhelpful.
Meri Beckwith Verified account @meribeckwith 6d
View quoted post on X
7 replies 13 reposts 105 likes 7.2K views View on X
Jacob Glanville Verified account @CurlyJungleJake
AI will accelerate some aspects of competitive intelligence gathering and portions of research, but not in the way that it is commonly hyped. The main limitation is that molecular interaction prediction, let alone kinetics of interaction prediction, is not remotely a solved computational problem. Biology hinges so extensively on binding kinetics that this precludes AI from “solving” biology simply with more compute. This is a fundamental difference between biology and math, computer science, or boilerplate drafting. There are ways forward but the road is much longer than many non-biologists would like to assume. wsj.com/tech/ai/silico…
www.wsj.com Silicon Valley Promises AI Will Cure Disease. Drugmaker CEOs Aren’t Buying It. No drug designed by artificial intelligence has won FDA clearance, and pharma leaders say biology is the obstacle.
3 replies 6 reposts 33 likes 2.2K views View on X