Sign in

Will Macnair

@willmacnair.bsky.social
1.3K followers 485 following 290 posts

Amateur neuroscientist, quite a bit of experience in single cell analysis, working on #neurodegeneration in Basel. Formerly postdoc with @markrobinsonca.bsky.social at UZH.

PostsRepliesMedia
Reposted by Will Macnair
ghost malone ergo propter bones @proptermalone.bsky.social · 30/09/2026
what i think is currently up in the AI space since we're having cost arguments again: selling inference is very profitable, but models are improving so fast that there's no time for capex on model training to amortize, so net at this rate they're losing money which is why they want a cartel
1533936
Reposted by Will Macnair
Masahiro Kanai @masakanai.bsky.social · 30/09/2026
Our single-nucleus immune multiome atlas is now out in @nature.com! 🧬 10M PBMCs from 1,108 @finngen.bsky.social donors recruited by Finnish Blood Service, profiled at @broadinstitute.org with chromatin accessibility and gene expression in the same nuclei, to trace how disease variants act 🧵👇
13317
Reposted by Will Macnair
Andy Matuschak @andymatuschak.org · 25/09/2026
Made a Printer Friend which quickly transcribes speech onto little slips of paper. Very nice for card sorts, quickly capturing open loops, discursive notes in books, etc. The speed of thermal printers and their auto-cutters is so satisfying! And dedicated mechanical key! Mm!
5555
Reposted by Will Macnair
Faine Greenwood @faineg.com · 25/09/2026
Imagine if instead of regulating drone risks the normal way (which we did), we instead paid weird bloggers lots of money to worry about superintelligent dogs sprouting thumbs and using drones to kill us. This is pretty much how the US is approaching AI risks and regulations. My latest:
little-flying-robots.ghost.io
Why We Can't Let Enormous Weirdos Regulate AI
Imagine if instead of regulating drones via the standard lengthy, boring, and bureaucratic process - which we did - we instead decided, in the name of fostering all-important American drone innovation...
25958279
Reposted by Will Macnair
Rob Patro @robp.bsky.social · 24/09/2026
If you are a US citizen or permanent resident looking for a postdoc, the President's Postdoctoral Fellowship Program is a great opportunity faculty.umd.edu/presidentspo...! If you are eligible, working in computational genomics, and interested in my lab, let me know!
faculty.umd.edu
President's Postdoctoral Fellowship Program | Office of Faculty Affairs
099
Reposted by Will Macnair
#SingleCell preprints @prepub-singlecell.bsky.social · 23/09/2026
Targeted single-nucleus sequencing of 39,800 neurons reveals extensive low-frequency somatic variants #SingleCell 🧪🧬🖥️ www.biorxiv.org/content/10.64898/20…
021
Reposted by Will Macnair
Harmit Singh Malik @harmitmalik.bsky.social · 24/09/2026
Some random thoughts: Good science doesn't always need disruptors. Often, it needs load-bearing walls. You can't pivot away from a failed multi-year experiment or "iterate" your way out of a mentee's PhD. Science needs stable foundations, not churn. Don't break the people building on top of it.
613022
Reposted by Will Macnair
Alex Kale @kalealex.bsky.social · 24/09/2026
What AI scales about science are the parts that are most broken and gameable. With the exception of combinatorial problems like protein folding, we need more thinking and less rote computation.
0236
Reposted by Will Macnair
The Transmitter @thetransmitter.bsky.social · 23/09/2026
A new publicly available single-cell atlas of the human prefrontal cortex was made from 6.3 million nuclei across 1,494 donors, ranging from before birth to more than 100 years. By @siddhantpusdekar.bsky.social www.thetransmitter.org/transcriptom...
thetransmitter.org
Prefrontal cortex has three transcriptomic acts
Various genes show distinct exits and entrances with age in the open- source dataset, the largest of its kind to date.
0124
Reposted by Will Macnair
The Transmitter @thetransmitter.bsky.social · 23/09/2026
Microglia don't just prune synapses; they also reach into axons and scoop out organelles containing cellular debris, a new preprint reports, calling it "skoupocytosis." #neuroskyence By @siddhantpusdekar.bsky.social www.thetransmitter.org/microglia/mi...
thetransmitter.org
Microglia scoop out garbage from inside neurons
Through a newly identified process called skoupocytosis, neurons recruit the immune cells to remove cellular waste they find hard to deal with themselves.
01910
Reposted by Will Macnair
papers-neurodegen.bsky.social @papers-neurodegen.bsky.social · 23/09/2026
Landmark map of human brain’s gene activity holds clues to Alzheimer’s disease and more #NeuroDegeneration 🧪🧠 www.nature.com/articles/d41586-026-…
021
Reposted by Will Macnair
papers-neurodegen.bsky.social @papers-neurodegen.bsky.social · 23/09/2026
Single-cell atlas of transcriptomic vulnerability across brain disorders #NeuroDegeneration 🧪🧠 www.nature.com/articles/s41586-025-…
021
Reposted by Will Macnair
papers-neurodegen.bsky.social @papers-neurodegen.bsky.social · 23/09/2026
Single-nucleus atlas of cell-type specific genetic regulation in the human brain #NeuroDegeneration 🧪🧠 www.nature.com/articles/s41588-026-…
021
Reposted by Will Macnair
Amanda Heidt @scattercushion.bsky.social · 23/09/2026
🧪Today, the PsychAD Consortium published EIGHT papers across Nature journals detailing their efforts to create the largest map of the human prefrontal cortex. It's hard to overstate the amount of work here, nor the absolute treasure trove this team has created for researchers.
nature.com
Landmark map of human brain’s gene activity holds clues to Alzheimer’s disease and more
The atlas of gene expression in the prefrontal cortex was created using samples from almost 1,500 donors of all ages — from infants to centenarians.
0119
Reposted by Will Macnair
Waggoner Lab @labwaggoner.bsky.social · 21/09/2026
Review @cellcellpress.bsky.social Oligodendrocytes in central nervous system health and disease www.cell.com/cell/fulltex...
084
Reposted by Will Macnair
Rob Patro @robp.bsky.social · 21/09/2026
It's here! new preprint: Gravlax, an annotation-independent molecular evidence archive for scRNA-seq. A count matrix freezes one annotation. Molecules never change. Instead, gravlax keeps the evidence. Requantify, query, and discover under any annotation. 🧵 www.biorxiv.org/content/10.6...
33413
Reposted by Will Macnair
John Gallagher @johnrgallagher.bsky.social · 21/09/2026
It’s very hard to fathom but we have intelligence without consciousness which our science fiction hasn’t prepared us for. Intentionless intelligence is weird.
6509
Reposted by Will Macnair
Ada Palmer @adapalmer.bsky.social · 20/09/2026
The world’s largest battery-electric aircraft flew for the first time on the 12th of August, a 27-minute test flight over upstate New York that cost about five dollars in electricity. Heart Aerospace’s 30-seat prototype is a big step toward a commercial version. buff.ly/i3f0PhR #ShareGoodNewsToo
techspot.com
The largest all-electric plane ever flew for 27 minutes, on $5 of electricity
Heart's X1 demonstrator flew for 27 minutes from a regional airport in upstate New York, running on battery power the entire time. The aircraft reached 1,100 feet...
191042293
Reposted by Will Macnair
Ed @ed3d.net · 20/09/2026
I mean that tbh. I block 10x the accounts I follow and a lot of times it's just "this person has an awful vibe" or "this person is a shitter in somebody else's mentions". The make-a-list types get INCREDIBLY mad with their "never interacted with me and blocked me" lists, but it's Good Actually
1863
Reposted by Will Macnair
Simon Willison @simonwillison.net · 18/09/2026
Being a computer scientist who refuses to find anything about LLMs interesting right now is a bit like being a geneticist who refuses to find anything interesting about the recently opened Jurassic Park
4181681
Reposted by Will Macnair
Per Engzell @pengzell.bsky.social · 18/09/2026
A paper is finished when the embarrassment of submitting it becomes smaller than the embarrassment of still working on it
630254
Reposted by Will Macnair
Omid V. Ebrahimi @omidvebrahimi.bsky.social · 17/09/2026
📜 Out now: ‘Towards a genetics of mental disorders as complex dynamical systems’ After 2 years of work with @michelnivard.bsky.social & @ludvigdb.bsky.social We outline a roadmap to align behavioural genetics with novel dynamic theories of mental disorders A thread 🧵 doi.org/10.31234/osf...
14719
Will Macnair @willmacnair.bsky.social · 17/09/2026
Super interesting new kind of model: - output is responses to structured queries you specify (e.g. Q = "which direction?", A = ["left", "right", "forward", "backward"]) - *extremely* fast Watch the doom demo here: typesafe.ai/blog/introdu... It feels like seeing the Terminator's internal state... 😅😬
typesafe.ai
Introducing System One Models & Jev - TypeSafe AI Blog
TypeSafe AI is an AI lab building machine-native intelligence infrastructure for automation, designed to make decisions within software. Try our first System One Model, Jev, in early access.
020
Reposted by Will Macnair
Richard McElreath 🐈‍⬛ @rmcelreath.bsky.social · 17/09/2026
Back in 2018 I needed a custom poisson-binomial log-prob for Stan, so I wrote a clunky one, and then Bob I think wrote a more efficient one. Today poisson-binomial is built into the math library. For all your unusual mixture needs. (My use was for weird survey data)
screenshot of poisson-binomial documentation. link: https://mc-stan.org/docs/functions-reference/bounded_discrete_distributions.html#poisson-binomial
1423
Reposted by Will Macnair
MC Stan @mc-stan.org · 17/09/2026
Release of CmdStan 2.40, including new Stan Math, core Stan, and Stanc, is now available! Blog post with new features blog.mc-stan.org/2026/09/16/r... Highlights - integrate_1d_double_exponential, a variadic version of old integrate_1d - integrate_1d_gauss_kronrod, a different quadrature rule ...
blog.mc-stan.org
Release of CmdStan 2.40
We are very happy to announce that the 2.40.0 release of CmdStan is now available on Github! As usual, the release of CmdStan is accompanied by new releases of Stan Math, core Stan, and Stanc3. Thi…
13016
Reposted by Will Macnair
Anil Seth @anilseth.bsky.social · 16/09/2026
1/6 An extraordinary @nature.com paper has just landed from the lab of @sergiuppasca.bsky.social. They found a way to create "xenocortical mice" which have ~92% of their cortex occupied by a cortical organoid derived from human stem cells.
69039
Reposted by Will Macnair
Rob Patro @robp.bsky.social · 16/09/2026
Our bramble preprint is out. Here's a thread about this work and why I think it's exciting! bsky.app/profile/robp...
092
Reposted by Will Macnair
Will Stancil @whstancil.bsky.social · 16/09/2026
No. This Cory Doctorow post is awful and borders on actual misinformation. The fact that a widely respected voice like him is spreading this kind of false narrative is, again, a giant blaring alarm about the effect the Bluesky info environment is having on people’s brains.
2329323
Reposted by Will Macnair
papers-neurodegen.bsky.social @papers-neurodegen.bsky.social · 16/09/2026
Inflection points and transitions in Alzheimer’s disease #NeuroDegeneration 🧪🧠 www.nature.com/articles/s41586-026-…
011
Reposted by Will Macnair
Tom Pepinsky @tompepinsky.com · 16/09/2026
the hardest thing for the politician-bureaucrats to understand -- but which NSF staff *absolutely do* understand and always have -- is that you actually make scientific progress by giving young people money and promising not to bother them
430084
Reposted by Will Macnair
Per Engzell @pengzell.bsky.social · 16/09/2026
Every dataset contains the answer to your question, provided you are willing to change the question
2659
Reposted by Will Macnair
Mattan S. Ben-Shachar @mattansb.msbstats.info · 14/09/2026
Upgrading my "hedghog-cake as an estimand" slide with some more hedgehog cakes from hell. #statsmeme
4588
Reposted by Will Macnair
Scott Barolo @sbarolo.bsky.social · 15/09/2026
🧵 Short thread: I give a seminar to 1st-year biomed PhD students in my program about how to critically read a paper, how to extract VAST amounts of information from the title, and why reading AI summaries instead of struggling with papers is *particularly* bad for trainees. 🧪
A comparative analysis of two paper titles, focusing what exactly we can expect to see in each paper. The verb choice alone shows that one is primarily a techniques paper.
28031
Reposted by Will Macnair
Sleegers Lab @sleegerslab.bsky.social · 15/09/2026
🧠🧬 New from our lab! Postdoc Celeste Laureyssen ( @celestelaureyssen.bsky.social ) reviews how neuropathology-centered GWAS can sharpen our understanding of the genetics of Alzheimer’s disease and its frequent co-pathologies—moving beyond clinical diagnosis alone. link.springer.com/article/10.1...
link.springer.com
A new era of genome-wide association studies in the field of Alzheimer’s disease and overlapping co-pathologies: lessons learned from a neuropathology-centered approach - Acta Neuropathologica
Alzheimer’s disease (AD) and overlapping pathologies represent a growing worldwide health concern. With the first disease-modifying treatments on the rise, it becomes increasingly important to move re...
055
Reposted by Will Macnair
Mr. Joshua @pantspants.bsky.social · 15/09/2026
Pen and watercolor. Two fools wander through an open field. The tall one dressed in a carrot costume and the short one dressed in an apple costume. Their hands are out to their sides to maintain balance. Carrot speaks, “You feel the ground shaking?” In the distance, we see a massive herd of wild horses thundering towards them. Hooves pound the ground racing to carrot and apple. By the time our boys realize their mistake it could be too late. Where are they going? Why wander dressed as horse treats? Things to consider. Artist’s note: Most of my drawings are black & white and I tend to use color sparingly. Hilary Knight, the illustrator of the Eloise books, is a big influence on both my linework and color use.
173646490
Reposted by Will Macnair
Zach Weinersmith @zachweinersmith.bsky.social · 15/09/2026
Can we please stop doing this? I made a comic about it and everything!
142240411
Reposted by Will Macnair
Matt Henderson @matthen.com · 14/09/2026
Chaos
827536
Reposted by Will Macnair
Sergey Ovchinnikov @sokrypton.org · 14/09/2026
Finally a more intuitaive way to learn pLDDT/pAE? 😎 sokrypton.github.io/protein_figh... (Character idea from @hannes-stark.bsky.social & Alex Waldherr)
727698
Reposted by Will Macnair
papers-neurodegen.bsky.social @papers-neurodegen.bsky.social · 14/09/2026
APOE-stratified genome-wide association analyses provide insights into the genetic etiology of Alzheimers’s disease #NeuroDegeneration 🧪🧠 www.nature.com/articles/s41588-026-…
011
Reposted by Will Macnair
Jieran Sun @onesinglecell.bsky.social · 13/09/2026
Our paper is out in Nature Methods! 🧑‍💻 There are 50+ methods for spatially-aware clustering on spatial transcriptomics data. How should we use them? A thread 🧵 (1/7) 📄 doi.org/10.1038/s415... #spatialtranscriptomics #bioinformatics #computationalbiology
doi.org
Beyond benchmarking: an expert-guided consensus approach to spatially aware clustering - Nature Methods
Benchmarking spatially aware clustering is challenging. SACCELERATOR is an open-source, extensible framework that formalizes expert-guided spatial clustering analysis.
1124
Reposted by Will Macnair
Chise @sailorrooscout.bsky.social · 11/09/2026
GOOD NEWS! A personalized mRNA vaccine has SUCCEEDED in keeping melanoma, a deadly skin cancer, from returning. Phase III trial results in MORE THAN 1,100 patients shows the vaccine SIGNIFICANTLY extended the time patients lived without their melanoma returning AND reduced the risk of metastasis.
statnews.com
Moderna and Merck say mRNA cancer vaccine succeeded in late-stage melanoma trial
A personalized mRNA cancer vaccine succeeded in a late-stage melanoma study, results that could herald a new, powerful approach in oncology.
4753111868
Reposted by Will Macnair
Dan Davies @dsquareddigest.bsky.social · 11/09/2026
As I keep pointing out, does no one understand Ricardian comparative advantage anymore. FACT: China produces over 1 million engineering graduates per year FACT: To date, China has not produced a single internationally marketable game show format Where does it look like our advantage lies?
37652182
Reposted by Will Macnair
Julien Roux @julienroux.bsky.social · 11/09/2026
Very good start of the #BaselSpatialOmicsSymposium today, with full auditorium and interesting talks! www.linkedin.com/posts/micros...
linkedin.com
#spatial | Microscopy Network Basel
Wow! It's amazing to see how the Basel Spatial Omics Symposium has grown in recent years. This year, we are holding the third edition. It initially started as a local Microscopy Network Basel event to...
161
Reposted by Will Macnair
Eris @isolyth.dev · 10/09/2026
New favorite painting discovered Symphony of the Sixth Blast Furnace (1979), Sedukhin
619527
Reposted by Will Macnair
🇵🇸Tim Henke (tɪm 'ɦɛŋ.kə) @timhenke.bsky.social · 10/09/2026
Prompted by the scandal around Navier-Stokes, Andreas Thom raises the possibility that OpenAI may also have stolen the work on non-sofic groups:
1/3 I want to record a number of points that I realized after reading up on the controversy. In fact it recalls an exchange I had with OpenAI after its non-sofic-group announcement. It exposes a serious, unanswered question about whether researchers can trust OpenAI with unpublished mathematics.

I wrote in an email to Mark Sellke and Sebastien Bubeck shortly after the surprising finding of a non-sofic group using the methods of Kun and myself:
“Another point is that I and a colleague in Dresden were discussing the expander matching problem and various extensions of the work with Gabor Kun actively over the last months with ChatGPT, so that we are of course curious if that was part of the training data or accessible to the reasoning process. There is a certain (frankly unacceptable) lack of transparency here; and I fear it will damage the communal process of math more than the new AI-generated results will benefit the subject.”

Mark Sellke’s complete answer was: “Regarding your conversations with ChatGPT: that did not happen.”

I had explicitly asked about two different things: (1) whether our conversations entered training data, and (2) whether they were accessible to the solving process. The categorical answer now looks as though it addressed only direct access under (2). No such qualification, explanation, or evidence was given. I take this as dishonesty to say the least.
OpenAI now says in the Buckmaster-Alpöge case that no specific user data was accessed, but adds that it “cannot rule out that de-identified data derived from their usage of our products helped improve our models.”2/3 OpenAI is drawing a distinction that its answer to me erased, despite the fact that my question explicitly made that distinction.
We are not required to reverse-engineer OpenAI’s internal training pipeline to establish what happened. Only OpenAI has the relevant data for that. If Sellke and Bubeck give a categorical denial, it must disclose the basis: product and privacy settings, relevant datasets and checkpoints, and what “de-identified data derived from usage” means.
I disabled model training on 29 June. That control is still only a promise whose implementation users cannot audit, and it is prospective: it does not answer what happened to earlier conversations or to derivatives already selected. OpenAI’s answer did not mention the setting or any account-specific check.
There is also a mathematical parallel, though a less explicit one. The approach of Kun and myself was not the main line of attack on non-soficity, in fact there were other more promising approaches along the line of quantum games etc. OpenAI’s detailed command of our techniques therefore made me wonder how the model found this route. If nonpublic research supplied by users improved a model and the provider then used that model to race those users to publication - without informed consent, disclosure, or credit - that would be ethically indefensible. De-identification may remove a name; it does not remove the intellectual content of a mathematical idea. Sellke and Bubeck seem to be blind to this simple moral aspect.3/3 Sellke’s categorical answer was, at minimum, unjustifiably broad and materially misleading; looking back it was plainly dishonest. Bubeck’s acknowledged attacks and his attempt to exclude Alpöge from a proposed OpenAI-proof paper deepen the concern that there is a loss of moral compass. Taken together, this raises serious questions about their judgment and personal integrity.
113748
Reposted by Will Macnair
Michel Nivard @michelnivard.bsky.social · 09/09/2026
It’s quite surreal how disconnected the AI discourse on the fans and critics sides is now… Very little sort of moderate appetite to do things like regulate mandatory water circulation, and green on/near site power generation for servers.
121
Reposted by Will Macnair
papers-neurodegen.bsky.social @papers-neurodegen.bsky.social · 08/09/2026
The hidden role of the bone marrow in Alzheimer’s disease #NeuroDegeneration 🧪🧠 www.nature.com/articles/s41593-026-…
011
Reposted by Will Macnair
papers-neurodegen.bsky.social @papers-neurodegen.bsky.social · 08/09/2026
Bone marrow myelopoiesis dysfunction in Alzheimer’s disease limits monocyte homing to the brain and drives disease progression #NeuroDegeneration 🧪🧠 www.nature.com/articles/s41593-026-…
011
Reposted by Will Macnair
Zach Weinersmith @zachweinersmith.bsky.social · 08/09/2026
For anyone not following the drama, there's been apparently a big development in ai math, along with some accusations of very bad behavior by OpenAI. But we don't have details yet. If I understand it, Buckmaster and Levent Alpoge made a big discovery on Navier-Stokes. Events go something like this:
17416157
Reposted by Will Macnair
Zach Weinersmith @zachweinersmith.bsky.social · 08/09/2026
You cannot "brute force" a 100 page proof of a theorem! You can convince yourself of this by counting the number of possible moves at each step, calculating the possible permutations, and comparing it to the number of atoms in the universe! It won't work.
1439128