The empty drops you threw out in your single-cell RNA sequencing analysis might be hiding mysterious things 👀 check out our bioRxiv preprint!
The empty drops do not contain cells. Yet we can still use them to learn interesting things about biology and technology. 1/
I tried to teach myself, but gave up after Christoffel symbols. Now I plead with ChatGPT to solve the ODEs in simple cases
Here is the "effective payline" for each institute, estimated (by Claude) as the percentile where one can expect 80% probability of funding from a logistic regression fit. The effective payline has gone from a historic ~12% to 6% in 2025.
wait the file suffix for cuda c headers is "cuh"? CUH!?!
*flashback* (i needed this rn)
kallisto sped up RNA-seq quantification by 50x. Now another 50x speedup... quantify hundreds of millions of reads in a few seconds.
This seems too good to be true. But it's true!
Incredible accomplishment by @pmelsted.bsky.social.
I'm sorry, 50 what now?
"We present a GPU implementation of kallisto for RNA-seq transcript quantification (...). For a large dataset of 295 million reads, runtime drops from 40 minutes to 50 seconds"
www.biorxiv.org/content/10.6...
good for you, man
wait what, i want to say when did this happen, but thats been answered
Similarly, if I dont inspect plans and verify implemention itself I can ask why it made certain decisions. It can point to rules, or at least say -something-. Prior to all of this it will provide multiple possible designs, with recommendations due to trade offs. I really recommend trying it.
I've been using a modified version of this -excellent- plugin system, which adds a plan-execute loop. Whenever an implementation plan is a bit wonky I can ask and it may point to a somewhat ambiguous rule/spec I defined. github.com/ed3dai/ed3d-...
"But it won’t know why it chose that path. It doesn’t feel friction and can’t explain if one architectural approach felt cleaner than another."
It may not truly know why but its postmortem can def provide clear reasoning for one decision vs another. Even for semi vague notions of the" clean".
🧬New from Evans et al!
📄A simple approach for multiple observations improves power to detect genetic effects and genomic prediction accuracy
👉 bit.ly/3PlkRA1
Super cool use of geometric autoencoder. I'm well convinced that we should all learn more about geometry to capture richer model landscapes enabling natural optimization techniques that bypass cumbersome constraints.
Problem is, it's just super hard
Very proud of this paper, now officially out at PRX LIFE journals.aps.org/prxlife/abst.... Here @mrazo.bsky.social shows that geometry-aware autoencoder architecture can discover low dimensional phenotypic spaces traversed by evolution. 1/2
Design PL designer and PL reviewer and pit them again one another
Wire clippings or sprinkles: the most dangerous game
Buddy I'm not entirely convinced IM conscious, let alone the programming machine
This is the meme. They're doing the meme.
No good can come from another round of “these nerds think AI is alive” discourse. Don’t engage. Save yourself. Eat some chocolate, watch your favorite movie. Do you realize how much more fun that is than yelling at strangers about a topic no one in the world understands?
The numpy constraint is so jax can immediately consume it for jax arrays
My "if I had more time" project would be a unified genotype to numpy toolkit in rust. Vcf/plink/bgen/plink2 -> numpy + polars (for sample/variant metadata)
Perturbation-guided mapping of colorectal cancer cell states to causal mechanisms https://www.biorxiv.org/content/10.64898/2026.03.03.708171v1
I fucking hate the American culture of entitled asshole selfishness. you see it everywhere. the driver assholes going 90 in a school zone. the assholes who don’t leash their dog or pick up their dog shit. the assholes who litter. the assholes who won’t vaccine their kids
Doing Things Right Is Tedious And Awful
meets
Thing That Does Not Comprehend Tedium
Apropos of nothing, ur reminder that high profile manu abstracts which emphasize model size over valid inference and predictions is a VC pitch wrapped in a veneer of academic prestige
California soldier among those killed in Iran war drone attack, Pentagon says
Also, happy endometriosis awareness month 🌻💛
People in my life have asked why I’m open about my health 🏥 and a BIG reason is to bring enough awareness to conditions to encourage ACTION ✊🏽
📣 There’ll also be events in New York City this week hosted by www.endofound.org if people are interested 💪🏽
dis is all u need www.biorxiv.org/content/10.1...
Great work by Albert from my lab! Albert has a few more papers coming shortly and will be graduating soon. Then off to a PD! He is a great PD catch and you should consider him :) bsky.app/profile/albe...
Or does there 🤔
