Lyle Ren

@DrSamuelBHume @moremyspeed Dementia isn’t irrelevant—same gene. DYRK1A doesn’t just degrade cyclin D1, it also phosphorylates tau and primes amyloidogenic APP processing, which is why AD pathology is near-universal in DS by the 40s… And Alzheimer’s & cancer are inversely related generally.

@moremyspeed I did say there’s an increased risk of leukemia despite the lower rate of solid tumors, and dementia/MR are not relevant in the context of cancer

Down syndrome lowers the risk of most solid tumors, and the molecular biology is fascinating. People with Down syndrome have 3 copies of chromosome 21, which carries a gene, Dyrk1a, that promotes degradation of a key cell cycle protein, cyclin D1 - the extra copy of Dyrk1a in Down syndrome lowers cyclin D1 levels, so the cell cycle slows down. This is one reason the risk of solid tumors is cut in half (the same is not true, unfortunately, of blood cancers, where the risk in Down syndrome is much higher).

@EricTopol @CellCellPress Interesting: plasma and CSF GFAP track two distinct astrocyte states — plasma is an early, amyloid-specific signal (Pereira 2021, cited), CSF correlates with microglial markers and doesn’t dissociate the same way (Benedet 2021). One green line in Fig 2 hides two biologies.

Stellar review of the recent and remarkable advances vs Alzheimer's disease (AD). Towards "a future where AD is not only treatable but also preventable." @CellCellPress cell.com/cell/fulltext/…

@DrSamuelBHume Wonder if GLP-1s modulate IL-11. I only see data on classical cytokines. But the hepatic fibrosis data point to a possible impact (as do maybe the Ozempic baby anecdotes).

This is cool: inhibition of the pro-inflammatory cytokine, IL-11, slows ovarian aging and preserves fertility (in mice and rats). This is the same mechanism that increases mouse lifespan:

@matthewherper Fascinating that Regev flags AI’s biggest risk as narcissism. Doesn’t that mirror our own very-human biases that kept us from innovating in Alzheimer’s, told us KRAS was undruggable (!), and laughed at Folkman’s theories about VEGF before Genentech changed the world of cancer?

Read my whole take here, and go ahead and ask me questions I've had a week to think about this. statnews.com/2026/07/06/ant…

So last week I spoke to @AnthropicAI CEO @DarioAmodei about how AI could change biotech. I also got to speak to other Anthropic execs, large pharmaceutical CEOs, and GLP-1 pioneer @lotte_bjerre. Here's what I learned: The most convincing thing to me is that Amodei recognized the limitations of biotech in way he didn't in his 2024 essay, “Machines of Loving Grace.” He forecasts that someday AI will allow researchers to make a decade's worth of progress every year. He said he's "very specifically not saying we’re going to get all this crazy stuff by 2028 or something. There’s simply no way to do it." I asked Amodei why we should believe in this tech/biotech revolution when past ones have failed. Pharma execs listed every reason to be cautious -- and why they'd be happy with 40%, 10%, or 5% gains -- and the nightmare scenarios that could make it all worse. But Lotte said this: "I would say AI is not going to replace scientists. I’m not worried about that at all. But scientists who are not using AI are going to get replaced."

@AnthropicAI Interpretability can bring insights to human biology. Ablating the J-space breaks the ability to bridge gaps, while immediate associations remain intact. This is the same split that makes delayed recall the earliest marker of dementia. Could inspire new early dx tests.

We invited experts in neuroscience, philosophy, and interpretability to share their perspectives on our work. Read their commentary here: www-cdn.anthropic.com/files/4zrzovbb…

New Anthropic research: A global workspace in language models. Of everything happening in your brain right now, only a tiny fraction is consciously accessible—thoughts you can describe, hold in mind, and reason with. We found a strikingly similar divide inside Claude.

@AmandaAskell Very true, and very challenging for patients (and also for doctors because frivolous lawsuits are absolutely an occupational hazard). That’s why it’s so imperative to free Fable to think beyond the classical medical silos and form a more integrated map of health and disease.

Extracting a probability from a doctor is one of life's unnecessary boss battles. Even if you beg them for an interval-valued subjective probability, and at that point you're basically asking for their hunch. I don't know if they get sued for giving out information or something.

@samuel_stanton_ THIS will be interesting. But even more interesting will be the white spaces you see. Science has a culture and the thinking becomes ingrained, which could be a blocker to wild and new ideas. Chasing amyloid is one major example; dismissing angiogenesis in cancer is too.

What is scientific taste? Tell us what you think in this 10-min Claude-led interview on how scientists choose problems, judge evidence, and decide what's worth doing. claude.ai/interviewer/ho…

@pineurosci This seems counterintuitive since psychedelics are often described as “opening” the brain, but what’s actually happening is they’re gating what reaches the DMN while shifting the directional balance from bottom-up toward top-down control. Explains utility in PTSD and autism.

New work on the effects of psychedelics on hierarchical cortical propagations out in PNAS! pnas.org/doi/10.1073/pn… Counter to our expectations, psychedelics attenuated bottom-up propagations into the Default Mode Network across substances and species. Thread below:

@AmandaAskell @ImmunoFever This is, unfortunately, a frequent lived experience especially for women. It’s also true that a lot of people suffer from doing “their own research” and using ineffective quack treatments. Just like powerful AI, powerful medicine needs a balance of access and guardrails.

I had chronic pain for most of my life until a doctor did an MRI of the pain source and found a congenital condition that was then fixed with surgery. Now I'm wondering if I had 30+ years of pain because doctors worried I was too stupid to be in the presence of scan results.

@C_Angermayer 🗿?

Major announcement coming this week Thursday from one of our longevity portfolio companies. It involves one of the most important pathways in aging biology. Stay tuned.

@IasonGabriel @GoogleDeepMind Fascinating paper and great frame with the waves! I wonder if multi-agenct coordination will end up being harder to solve than intelligence (CooperBench coordination challenges.)

AGI is sometimes framed as a single end point; more likely are successive "waves" of technological transformation... 🌊🌊🌊🌊 For more on this topic—and the possibility of Artificial Superintelligence—check out this @GoogleDeepMind report, led by Tim Genewein!

@IdoAizenbud This week’s @ShorterLab preprint makes your finding architecturally beautiful: axonal mito lack mtDNA and don’t produce ATP; dendritic mito do. NMDA computation and OXPHOS power are co-localized in dendrites. The axon has neither. The brain solved the von Neumann bottleneck.

What can a neuron compute? Real biological neurons are complex, but how capable are they? Using a new method, we found that a single cortical neuron can classify cats vs dogs, recognize spoken words, and solve 10-bit parity, all tasks thought to require entire networks. (1/15)

@GrantHBrennerMD @dspetro TMS cascading into different spheres is two plasticity dimensions at once. Treat the circuit rigidity (OCD) and the structural depletion (MDD) starts to move. I wonder whether a single intervention can do enough of both, or whether this is where psychedelics come in.

@dspetro That as well. But I was specifically thinking of the multiple layers with OCD and MDD given treating depression with TMS, in particular, and seeing the subsequent neuroplastic changes unfolding afterward in different spheres when it is effective.

Obsessive compulsive disorder and major depressive disorder together can be very recalcitrant. Ocd can lock in a lot of rigidity and repetition behavior and make change difficult, and the clinical depression can feed into obsessional negative self thoughts and models of being unable to change. The behavioral piece and the interpersonal impact can make it so that the person's whole life almost becomes dependent on staying clinically depressed. Particularly when other people's ways of thinking about the person have become established and more so if those others have come to find purpose in being a caregiver. In many cases the whole family system and generational modeling is part of the solution - the social aspect. The presence or absence of insight appears to be a key factor but may at times be more phenomenological than causal. And the clinical depression again is often synergistic. Finding the key sequences so to speak which can untie that gordian knot...

@ImmunoFever Wow! Perhaps the same evolutionary logic as RBCs: strip the machinery that pollutes. RBCs don’t burn their own O₂. Axonal mito don’t generate ROS along their own transmission line. Glycolysis does the work. Dirty energy stays in the soma. The axon stays clean — by design.

We need to rewrite the textbooks!!! This blows my mind🤯! Instead of what we always assumed about mitochondria in the axons producing energy, NOPE. Backwards! They're actually using energy to build a gradient. I'm wondering if this is analogous to how we make red blood cells. 1/2

@mgdurrant @parmita @Abhinavsns Thank you! One comment: limiting to institutions reinforces the status quo; hopefully independent researchers will be able to access soon (with appropriate guardrails).

@parmita @Abhinavsns claude.com/form/mythos-ac…

Mythos is an excellent biologist. After we first gained access to it, we tested its ability to perform agentic molecular biology research and propose new hypotheses. It was a significant improvement, its biological reasoning and taste are impressive. We give more examples here:

@NatureMedicine Interesting! Oligo_1 appears in TD2 — the earliest transition, before significant tau accumulation. Myelin isn’t just a bystander. It may be loading microglia with lipid before the plaques fully form. This shows when the switch flips. Is the oligodendrocyte who’s flipping it?

Microglia cellular changes underlie divergent resilience-associated mechanisms in human aging. dlvr.it/TStFqw

@AnthropicAI @OmicsOmicsBlog Love “like driving through cities built before cars.” And in psychiatry, we’re also navigating without maps. Oncology has hallmarks — shared signage, agreed territories. It’s why cancer medicine accelerated. Psych is still routing via crumpled napkin, drawn by committee in 1994

New Science Blog: Why has AI advanced faster in coding than in biology? To agents, bio databases are like cities built before cars—maddening to drive in because they're designed for different traffic. How do we build infrastructure agents can use? anthropic.com/research/agent…

@awaisaftab Brilliant piece. The two axes tell us where each brain has landed, not why. GWAS by biotype and onset timing both seem like missing keys — especially for BT3, where perimenopausal vs. adolescent onset may be the same cluster for completely different reasons.

@_Hao_Zhu @StanfordHAI These seem to be the exact failure points as human teams coordinating with poor organizational governance. Are there org design principles that could be borrowed from management science to course-correct?

> If we are to realize the promised future where AI agents work together – and with human collaborators – AI will have to be good collaborators, but so far they appear to be lone wolves Our research on the curse of coordination in agent teams featured in @StanfordHAI blog hai.stanford.edu/news/ai-coding…

@SriniGRao @C_Angermayer @biospace This is an amazing development! The missing link to implementation is helping prescribers match the condition to the compound, and for that we need to build a framework for how to think about therapies. Here’s something we’re working on:

Thankful to take part in this thoughtful @BioSpace discussion on the potential opportunities - and real implications - of last week's Executive Order on psychedelic medicine. Policy signals matter, and how they’re implemented will shape innovation, timelines, and ultimately patient access to new options for difficult‑to‑treat neuropsychiatric conditions. Link: biospace.com/drug-developme…