Prof. of Neurobio & BioE @Stanford @MichaelLinLab ☘️🧪🦠🧠🔬📖🇺🇲🌏 Neurons, viruses, medicines. Inventor of the Paxlovid predecessor. Bots/bad manners blocked
Harvard→UCLA→HMS→UCSD→Stanford
- Tweets 18,288
- Following 464
- Followers 29,077
- Likes 9,172
Pinned Tweet
Click the thread below to get to my SARS-CoV-2 mega-metathread.
That thread previously had its own intro tweet, but I deleted it by mistake. This being Twitter, there are no free undos. So instead this tweet will parachute you into it.
16
17
132
139,540
Michael Lin, MD PhD 🧬 retweeted
Everyone who works in “AI or Techbio” should be forced to shadow in a lab for at least 1-4 weeks to see what the next steps after you have a structure on your computer.
And that’s just boring brute force research, not exploratory research with new hypothesis.
there will be within the next few years stunning advances in biology. there will whole new ways to think about what biological study even means. i would be only mildly surprised if there were near term solutions to problems like cancer, neurodegenerative disease, etc
13
24
231
7,111
Michael Lin, MD PhD 🧬 retweeted
When people start to write postmortems of the GenAI crash, this will be a popular statistic:
venturebeat.com/infrastructu…
25
122
708
17,492
This week feels like living in a bizarro world:
Losing money is celebrated; profitability questioned.
Ignorance is trusted; knowledge dismissed.
Criminals are defended; victims attacked.
Cowards attract followers; the courageous are shunned.
2
1
13
1,290
I mean, I can say "patience is overrated" without also saying honesty, knowledge, experience, and productivity are also overrated. But we're seeing the bizarro world spreading to multiple value judgements.
1
1
246
In case you needed any evidence that there’s a bubble in AI.
So the hot thing is starting a drug company with no knowledge data patents or experience. But hey they have a computer.
A very very common bio startup idea I've noticed people converging on lately: The Neo Pharma
The Neo Pharma is a startup that develops a pipeline of drugs (agnostic to disease/modality) with an underlying AI system around asset selection, program management, clinical operations, etc. Most common is the idea to use AI to capture and systematize drug hunter expertise
Unlike a biotech, where key innovation lies in the technology, the Neo Pharma is differentiated by its internal operations with AI
The catch for AI/tech people entering bio: You only find out if you're differentiated ENOUGH in 7+ years, based on initial drug portfolio & success rate against regular Pharma (and all the other Neo Pharmas in your cohort)
It's an obvious, ripe idea for biotech in the age of AI given most biotech work happens in email inboxes already (directing overseas labs to do the physical experiments, manufacturing, trials). It's also a bet that AI can shift the underlying economics of drug development
Curious how much in-house AI models will matter for these co's. Is it worth trying to aggregate proprietary data around specific biopharma workflows OR will this aspect be democratized soon enough and execution/capital is the only variable factor?
14
19
353
24,730
Michael Lin, MD PhD 🧬 retweeted
Feels like bad sci-fi movies becoming reality..
Warhound, the brain child of Ben Lamm and Palmer Luckey, emerges from stealth to genetically engineer dogs for military and law enforcement.
spr.ly/6011BHZNMF
2
2
10
2,327
Michael Lin, MD PhD 🧬 retweeted
Not the only reason. Chinese VCs invest in very early stage companies. Where as many US VCs will not look at a company until it is at DC or Ph 1, Chinese VCs will write small checks to founders with just an idea.
China has gotten ahead of the US in biotechnology, largely by making their clinical trial system more effective. I covered this in a @nytimes article.
Now, American patients are travelling there to have access to drugs they can't access at home. It's high time the US woke up and implemented much needed reforms.
12
10
66
6,088
Michael Lin, MD PhD 🧬 retweeted
This is a reasonable first step towards keeping the center of biotech IP generation (and value creation) in the US
I’m sure we’ll see complaints and Peter will advocate “eurowashing” as a workaround
Next, we should focus on adjusting reimbursement to motivate US discovery & dev
The regs are out: Prohibited tech includes pharmaceutical compounds, including a number of important modalities as examples
1
3
6
1,561
Where did 2025’s 30 best-selling medicines actually come from? Industry made more of the molecules than the “pharma just buys innovation” story suggests, but 28 of the 30 rest on biology first found in academic labs. Where a molecule starts tells you surprisingly little. Maybe an interesting set-up for AI and the next 15 years...
asymmetriclearning.substack.…
9
12
95
10,884
Some protection from a boost is better than no protection, and it looks like there will be high rates of infection soon
Some more info:
smithsonianmag.com/smart-new…
2
3
353
PSA: get your flu shot now if you haven't already
Using last year's flu trends as a template, from today we expect infection rates to double every 1.5 weeks, peaking ~2 months from now.
After vax, immunity takes 1.5 weeks to kick in, peaks at 1-2 months, decays after 3 months.
1
3
25
2,313
Influenza is ramping up early due to G155E, which partly evades this year's flu shots, even though the shots did get the strain right: H1N1.
(Jesse Bloom predicted this last month, not surprisingly. A reminder that lab work can predict disease trends)
For H1N1, many new natural strains have modestly reduced neutralization by human sera relative to strain in 2026-2027 Northern Hemisphere vaccine. Strains w lowest neutralization often have G155E, which has emerged repeatedly in natural strains, especially in D.3.1.1 subclade.
1
1
7
651
Michael Lin, MD PhD 🧬 retweeted
Ignorance isn't the biggest problem.
Confidence isn't the biggest problem.
It's ignorance combined with absolute confidence.
18
41
303
4,172
Michael Lin, MD PhD 🧬 retweeted
My theory about corporate AI psychosis is that AI is most appealing to people who aren’t good at their jobs. AI slop looks incredible to them because they’re so incompetent they don’t even know what good work looks like in the first place.
105
1,762
14,216
211,997
Nvidia CEO Jensen Huang has endorsed California's potential new billionaire tax laws, even if it will cost him $8 billion.
"I don't mind paying taxes. The more you make, the more taxes you pay. I'm okay with that. I am the ultimate American Dream."
509
1,915
37,044
2,187,468
Michael Lin, MD PhD 🧬 retweeted
Not only is this idea old, I’ve seen it in department seminar. Monkeying with lipid transport is almost as old as the field itself. It didn’t exactly escape notice that the most well known risk factor for AD is a lipid/cholesterol transport molecule…
GPT-Rosalind surfaced a new lead for drug discovery: could restoring cholesterol transport protect neurons from degeneration?
DCTN1 is a gene linked to Perry syndrome, a rare neurodegenerative disorder. In mouse-brain data, disrupting it increased expression of genes involved in making cholesterol. GPT-Rosalind suggested a possible explanation: cholesterol may not be reaching where neurons need it.
That gives scientists a specific mechanism to test and, if confirmed, a direction for developing new treatments for neurodegenerative disorders.
With @perturbai_tx and @scrippsresearch, a team of thousands of GPT-Rosalind agents worked in parallel, connecting gene-expression results from millions of mouse brain cells with scientific literature in just 32 minutes. We’re now expanding this collaboration to identify new drug targets.
openai.com/index/scaling-ai-…
8
6
185
12,430