Finished listening to this excellent episode. Great exercise in triangulation of different lines of evidence! So should you be taking statins?
Maybe! Might make sense to get your cholersterol levels checked!
*Any weekend plan*
Me:
Main goal of podcast achieved ✅
Hope you enjoy the new episode! open.spotify.com/episode/792E...
Me and Jacob holding ice creams and smiling
Wooohooo, our podcast Hard Drugs now has listeners in 100 countries! 🥳
Now's the time to share that when we brainstorming the name, @jacobtref.bsky.social suggested adding the subtitle 'by two softies' 😆
*30 minutes in*
**....people infected with the virus have about 500 quadrillion particles of viral surface proteins circulating in the antibodies during an infection.**
Jacob: That's...that's a lot. I am no mathematician. 🤣
If you have a long drive or flight coming up, I recommend our latest Hard Drugs episode!
The story of developing the hep B vaccine is wild, and it was very fun to record. This episode also contains the most swearing.
So far this is the Hard Drugs episode I personally learned most from while recording. Some holiday listening for those of you traveling this week (I’m headed to the airport in 30 mins…).
Fun to record 8 episodes this year. With that, signing off for 2025! ✌️
screenshot of my post
Big new blogpost!
My guide to data visualization, which includes a very long table of contents, tons of charts, and more.
--> Why data visualization matters and how to make charts more effective, clear, transparent, and sometimes, beautiful.
www.scientificdiscovery.dev/p/salonis-gu...
I appreciated the reflection on how insane 18th and 19th century science seemed. What do you mean you put a dog's brain into a rabbit's brain? Wait, you are carefully observing a sealed dead eel?
THIS is what I need no more ryan lizza and so forth etc
Attenuation Is All You Need (Well, inactivation too)
We cover the invention of 20 vaccines from the 1700s through 1970. Technology (smallpox vaccine) preceding scientific understanding (germ theory), then speeding up once theory established from experiments (maggots…) & new tools made (agar plates)
Some news: Open Philanthropy is now Coefficient Giving! Our mission is unchanged but the new name reflects our growing work with other donors to multiply the impact of their giving.
🧵 on our work to make philanthropy a more efficient "market" and plans going forward:
Some amazing jobs here! Our science and global health R&D team works on funding breakthroughs for neglected, high mortality/morbidity diseases and does a ton of great and interesting work -- see job links in Jacob's thread!
I blogged more about the roles here: blog.jacobtrefethen.com/hiring/
Please share the job descriptions with colleagues and friends who you think may be interested! 🙏
Screenshot of Open Philanthropy blog post on 5 science giving highlights
You can get a sense of what we support from this piece last year. We fund whatever we believe is important, neglected, and tractable across many areas of discovery and translation, which leads to a broad portfolio. Generalists very much welcome!
www.openphilanthropy.org/research/fiv...
Hiring a PO to help us give away tens of millions more in vaccine R&D over the coming years, in particular to make progress on strep A. Kills 600k/y but should be fixable with creativity...
Looking for an experienced scientist, but not necessarily in strep A:
jobs.ashbyhq.com/openphilanth...
Senior Program Associates will work closely with our existing (world class!) POs to help scale grantmaking
Variety of research & analysis, interacting with grantees, portfolio management
Fun for people with technical backgrounds who are impact focused:
jobs.ashbyhq.com/openphilanth...
🚨 Hiring scientists to help give away money 🚨
At Open Philanthropy, we have given away $600M in biomedical research since 2016. Now, we're expanding the team to do more.
We're hiring:
* Up to 3 Senior Program Associates
* Program Officer in strep A
Remote OK, salaries in links! (Thread below)
This one is kind of for you tbh, made us happy to think of your reaction if we dropped it
And then if that all looks good, you're going to submit a huge data package to the FDA and say, "Can I please sell this drug in America?" The FDA will take 6 to 10 months and review your data, review your thousands of pages of submission and get back to you with a thumbs up or a thumbs down. After you're selling your drug, you're still collecting data. The FDA might require further studies after they approve your drug if there are particular questions they have that things should be addressed. If in those studies you end up with a negative result, they might withdraw your ability to sell the drug; that happens somewhat frequently. Also, you're going to be collecting in the real world, more side effect data. Once hundreds of thousands of people are using a drug, you will spot more side effects. They won't be randomized, so you won't necessarily get as high quality data, but you at least get more data as things come in, so the evidence collection does not stop once you get approval. Saloni Dattani: Wow, great. That's a very long process. So I have two things about mice that I wanted to talk about.
MFW my co-host @jacobtref.bsky.social is telling me all about the drug development pipeline.
Spotify likes to do a second transcript of its own, and gave you a new name:
Welp @scientificdiscovery.dev and Jacob are back again, Will AI Cure Disease? By the end of the 4.5 hours you'll be significantly older and might find out!
www.worksinprogress.news/p/will-ai-so...?
Will AI solve medicine?
We decided to be as definitive as is possible in 2025. That meant going long, through the drug development process
Sections
1. Clashing worldviews
2. Drug discovery
3. Models
4. Efficacy
5. Safety
6. Manufacturing & healthcare
7. Funding
8. Trust & ambition
Hope you enjoy!
YouTube comment that says "Fascinating! Almost didn't realise this was AI generated content until I noticed the blink rate and ran it through a model."
Apparently some people think I'm AI generated, because... I don't blink enough
Tune in to find out:
- how to go from hallucinated cat pictures to hallucinated proteins
- whether @jacobtref.bsky.social can do photosynthesis
- whether we can make tiny protein straws for tardigrades to drink from
Our most practical episode yet for people who want to dabble in biotech
As of 2022, you can hallucinate protein structures using AI similar to Midjourney (RFDiffusion) -> create amino acid strings for them (ProteinMPNN) -> validate with AlphaFold
Biology becomes engineering…
Cool episode on how AI can help us design new proteins (e.g., for vaccines).
I’d heard of AlphaFold before, but not of ProteinMPNN. You feed a protein structure into PMPNN & generate possible sequences of amino acids. And then you feed those into AlphaFold to check whether they’d fold up that way!
me too!
Hearing a lot about AI designing drugs, but not sure what specifically that means?
Ever wonder how AlphaFold works, or what practical problems it helps with?
How rattan daybeds can hold their own in a modern home?
This episode is for you!
every time i feel bad about writing long essays / blog posts i think, no. what would saloni and jacob do.