T cells do the heavy lifting when it comes to the anti-tumor action of checkpoint immunotherapy. But do antibodies play a role too? That’s the question @yile_dai looked to answer in his thesis work, out today in @Nature! 🧵below nature.com/articles/s41586-0…

I encourage protein design tool developers to consider first-class support for apple silicon. My experience with BindCraft has been eye-opening. With a little kernel engineering, an M5 max MBP can hold its own against dedicated NVIDIA cards. Time per full trajectory shown below.

Weekend project complete: Run (Free)BindCraft on your mac! Check out the latest version 1.0.6 which supports experimental support for apple silicon (github.com/cytokineking/Free…).

We're expanding access to GPT-Rosalind for academic research. The new Rosalind Academic Plan helps scientific labs bring AI into the work of understanding biology, exploring ideas, and deciding what to test next—with sponsored access for eligible labs at U.S. universities. Academic research starts with questions whose impact may take years to impact human health.

RootPath is a really strong outfit. We have worked with them before on complex DNA library synthesis and they have collaborated with top groups in the TCR space. So I'm excited to see this new service, we will be trying it out.

Looks like a great opportunity given the early strong vibes with Gemini 4 Argon

If you don't think that there is a bubble in AI x Bio valuations, just consider that the purported valuation of Isomorphic Labs is more than that of Revolution Medicines. Investors are acting as if preclinical Isomorphic already has an approved KRAS scale blockbuster.

The Ariax agent is definitely the easiest and now possibly the cheapest way to generate binders!! And it has BindCraft2. Anthropic competition enjoyers take note!

I am shocked, SHOCKED I tell you that OpenAI is changing their subscription model after I have been bragging about how I abuse their generous usage allowances, circumvent limits, and signed up for 5+ max accounts to run my slopbots and never-released "apps."

We’re honored to be named to the 2026 #Endpoints11. This recognition reflects the support of our investors and Board and we look forward to the work ahead as we advance STP-100 toward the clinic in early 2027. Read more: bit.ly/3TLPq4F

Anthropic: it's too dangerous for you to apply our open-source speedups to your open-source pipeline.

Huge congrats to @MartinPacesa and team for the first big release from his lab. BindCraft2 is absolutely 🔥. The speedups are amazing and there are so many added features. Plus it's completely free for academic and commercial use. 🐐

Anthropic has been absolutely cooking in protein design with big contributions to open-source models. How amazing would it be if @nc_frey and his crew dropped an MIT-licensed SOTA antibody design model? Can't think of a bigger accelerator to democratizing drug discovery.

Starting today, teams and institutions can apply to our Life Sciences Verification Program (LSVP), which gives verified researchers and organizations access to our most capable models for professional biology and drug development work. LSVP is built for teams of all kinds: academic labs, startups, non-profits, biotechs, pharma companies, and more. Learn more and apply here: claude.com/form/life-science… We know that it has been frustrating to encounter blocks on Claude Fable. LSVP is our mechanism to prevent biological misuse, like the documented cases we’ve disrupted, while enabling legitimate life sciences research. LSVP includes a new set of safeguards to provide a better UX for life sciences researchers. We believe that advancing biology research and drug discovery will realize some of the most important beneficial impacts of AI. Please DM me if you run into any issues with the application.

Congrats to Alex and Ben for their $40M(!) grant from OpenAI Foundation to work on cancer vaccines! A million data flowers will bloom.

Hi all, The Lp(a) HORIZON trial has released topline data and, quite shockingly, missed its primary endpoint. In other words, lowering Lp(a) in patients with prior MI, stroke or peripheral arterial disease, who were otherwise very well treated for LDL-C, blood pressure, diabetes and other risk factors, did not reduce the primary cardiovascular endpoint. We obviously need to see the full data before making firm conclusions, and I don’t want to speculate too much without the details. But this is a big enough result that it is worth summarizing what we know, what we don’t know, and what this may mean for our patients after 20+ years of trying to test the “Lp(a) hypothesis.” What we know: 1-There are hundreds if not thousands of genetic, epidemiologic and Mendelian-randomization studies showing that elevated Lp(a) is associated with MI, stroke, peripheral arterial disease and aortic stenosis. That body of evidence is very strong. 2-However, much of those data come from community-based populations, often before the era of intensive LDL-C lowering and modern secondary prevention. 3-There has been much less information about how much residual risk Lp(a) carries in someone who has already had an event and is then treated very aggressively. 4-HORIZON may have had some of the best-treated patients of any recent cardiovascular outcomes trial. Baseline LDL-C was about 65 mg/dL, a measured LDL-C contains the cholesterol carried on Lp(a), so in reality, 15-20 points lower. 5-In patients with very high Lp(a), if you correct LDL-C for Lp(a)-cholesterol, the actual LDL-C carried by LDL particles may have been closer to 45–50 mg/dL, perhaps even lower in some patients. 6-This raises a very basic question: Can you still demonstrate a major incremental benefit from lowering another apoB-containing particle when the underlying LDL burden has already been driven this low? What we don’t know: 1-What was the actual corrected LDL-C in these patients? I think it would be extremely informative to directly measure Lp(a)-C and calculate corrected LDL-C. This may tell us a lot about the biological setting in which pelacarsen was being tested. 2- What was the OxPL status? Our prior work has suggested that much of the pro-inflammatory biology associated with Lp(a) is related to its enrichment in oxidized phospholipids. Did OxPL fall? Did patients with higher OxPL derive more benefit? Was Lp(a) concentration actually identifying the patients with the most pathogenic particles? 3- Did we measure the right component of Lp(a) for trial inclusion? We generally measure molar particle concentration. But is molar concentration itself the main driver of risk, or is it partly a surrogate for what the particle carries? Cholesterol? Triglycerides? Oxidized phospholipids? Other proteins? Could two patients with the same Lp(a) concentration have very different Lp(a)-mediated risk? I think this question deserves much more attention. 3- Were the genetic data telling us exactly what we thought they were telling us? The genetic data are extremely compelling, but genetics reflect lifelong exposure. A clinical trial treats patients late in life, often after decades of arterial injury and after an event has already occurred. Those are not necessarily the same experiment. Could there also be some unrecognized biology linked to the LPA locus that we have not completely accounted for? That possibility should at least be considered. 3- Does very low LDL-C modify the Lp(a) risk relationship? Maybe Lp(a) is particularly important when LDL-C is higher, but its contribution becomes smaller once LDL-C is driven to very low levels. Again, we need the data. 4- Does aspirin or other antiplatelet therapy reduce part of the risk associated with Lp(a)? Lp(a) has potentially important prothrombotic effects. Almost everyone in a trial like HORIZON is receiving contemporary antiplatelet therapy. Could that blunt one component of the risk associated with Lp(a)? 5- Why are these patients still having events? This may be one of the most interesting questions of all. These are patients with LDL-C around 65 mg/dL, and perhaps corrected LDL-C substantially lower, yet cardiovascular events continue to occur. What is driving that residual risk? Inflammation? Thrombosis? Plaque burden that is already too advanced? Other lipoprotein characteristics? Something we are not measuring? 6- Do we need to re-examine some basic assumptions about atherosclerosis? We have spent decades focusing heavily on the quantity of circulating lipoproteins. But perhaps lipoproteins are relatively benign until they undergo biological modification in the artery wall. Oxidation may be one of those key modifications. For some patients, the answer may be to remove more particles from the circulation. For others, perhaps the better approach is to prevent their oxidation or block the downstream biological effects of oxidized lipids. The recent difficulties with anti-inflammatory approaches, including IL-6 inhibition, make these mechanistic questions even more interesting. 7- Was there something specific about pelacarsen, the degree or timing of Lp(a) lowering, advanced disease, trial duration, background therapy or patient selection that mitigated a potential benefit? We simply don’t know yet. That is why the detailed results will be so important. What does this mean for patients today? If you have already had an MI, stroke or PAD, the immediate lesson is very clear: 1- Get all of your established risk factors treated aggressively. 2- Get LDL-C/apoB very low. 3- Control blood pressure. 4- Control diabetes. 5- Don’t smoke. 6- Use appropriate antiplatelet and other guideline-directed therapies. HORIZON shows us what modern secondary prevention should look like. If you have elevated Lp(a) but have never had an event, the genetic and epidemiologic data still suggest increased lifetime risk. Until the other 4 outcome trials read out, I would continue to treat every modifiable risk factor aggressively. We should wait for those trials before drawing broad conclusions about the entire field. I think the story of Lp(a) therapy is beginning, not ending. We also need to show tremendous respect and gratitude to the patients who participated in HORIZON and to the investigators and companies that invested enormous resources to actually test the Lp(a) hypothesis, to the ultimate benefit to peole with elevated Lp(a) to best guide how to manage risk. More to come as we go forward.

Very informative from Sam- and honest too!

I'm absolutely shocked by the Novartis LP(a) ASO flop. The genetic and epidemiological data for this approach could not have been more of a slam dunk. Are there any smart cardiologist/lipidologists out there who can possibly explain why this failed? Biology is humbling.

Lilly's acquisition of Merida is remarkable. You gotta love seeing big pharma take bets on innovative science like this. Feels like we are about to see some big battles between 'immune reset' and precision autoantibody deletion. Patients win either way!

We previously showed that APS-1/APECED features excess IFNg responses that can be targeted therapeutically (Oikonomou NEJM 2024). We now extend this to show that early JAK inhibition can intercept the development of irreversible autoimmune endocrinopathies nature.com/articles/s41467-0…