Rob Patro

My thought on AI review: lh3.github.io/2026/10/27/l... TL;DR: journals are banning fully AI-generated review and I think that is the right decision

Elon is, indeed, dumb enough to believe that Democrats would hold people responsible for their illegal, unconstitutional, and often criminal actions. If they regain power, as soon as they do, they'll focus on some nonsense "unity" narrative, and fail to act until they fumble and lose again.

“Professor, I’m not sure I recognize this dialect. Is it redditor?” “No, but you’re close. 21st century, yes. It’s late-web AI denialist. Like redditor, AI denialist was consumed with shibboleths and catechisms. But instead of narwhals and bacon, it’s parrots and plagiarism machines.”

They don't have arguments, only tribe, only purity tests.

Because the same mentality that leads you to ban all AI assisted content, regardless of what it is, leads you to ban non-english posts. This is who the people are who align with their tribe and refuse to think.

Yup

You do not hate C++ enough

IMO, if C++ really fails to ship contracts here, thereby slipping them to C++29 at the earliest, it’s another nail in the coffin and C++ will continue on its slow decline toward being a legacy language.

"How do we kill vibe coding and AI slop?" Vibe coding and AI slop are bad. Applying these terms dishonestly or lazily is also bad (though not in the same way). combine-lab.github.io/blog/2026/10...

Emily Bender going around dumping on people who are saying "AI isn't good for this task" because they aren't being anti-AI enough 🤣 Like Bill says, it's a Temperance movement

Oh and the headline is 26-83x faster than minimap2 for transcriptome mapping (we're not doing spliced genomic mapping yet) with memory usage ~2.1G and similar accuracy (and equal or better quant so far). Feeling cute; may even eek out a little bit more before the preprint.

Dear #bioconda maintainers; I’d appreciate a merge when you have a chance! github.com/bioconda/bio...

For the morning crowd!

👀 — new fast long read transcriptome mapper, who dis? crates.io/crates/spear... Lots more details to come, but there are some docs in the meantime: combine-lab.github.io/spearfish/ Oarfish support incoming.

Yes. Reviewers should stop including benchmarks with the nvidia RTX5090. For all intents and purposes, it is no longer a consumer gaming GPU and so doesn’t belong in consumer gaming benchmarks. youtu.be/ZgyL-6jLEwo?...

The Presidential postdoc program at UMD is accepting applications through Nov. 1. If you are a US citizen or permanent resident, please consider applying. If you're interested in working with my lab through this opportunity, let me know! faculty.umd.edu/presidentspo...

It turns out that the polynomial 2-OPT approximation to SCS released a few weeks ago by OpenAI can actually be made quite efficient in practice. I have an implementation here github.com/rob-p/practi.... It would be interesting to see if there's a well-known instance set to measure on.

While people are discussing the hazards of AI generated content on arXiv... arxiv.org/abs/2610.29766

Uhhhh @biorxivpreprint.bsky.social, are you OK? Does this have anything to do with 4?

For the morning crowd!

This article is making the rounds, and it's good to have a discussion on this topic. Number 2, in particular, seems a very tenuous and suspicious claim. I don't see why it would hold true, and it certainly doesn't seem justified.

My department (UMD Computer Science) is hiring! If you're on the market, you should apply. LMK if you have any questions not answered in the ad. umd.wd1.myworkdayjobs.com/en-US/UMCP/d...

Turns out the order of your scRNA-seq QC pipeline matters more than we thought: removing ambient RNA before doublet detection can reduce doublet detection performance! Our new paper from the amazing PhD student @michellemeier.bsky.social

New journal idea just dropped: "Journal *actually for* Computational Biology" Premise, we should let people actually access the articles and read them, even if they are *only* 31 years old.

Order Matters: Ambient RNA correction confounds doublet detection in single-cell RNA sequencing https://www.biorxiv.org/content/10.64898/2026.10.13.758369v1

Sony & their pro DRM nonsense represent the worst of enshittification. I hope the jailbreak community is successful in cracking open their platform, and allowing people to fully use the hardware they bought at great cost Sony blocks PSN access for unupdated PS5s share.google/rVhOp9SpxPQM...

I agree with Stephen on the main point here. I think there may even be reasons to go even further, but certainly, as a practical matter, I don't think such disclosures are long for this world. I also think they have some coherence problems from first principles.

What’s so nice about this is that it’s like an independent, better, and more generic version of the CellRanger chemistry auto detection. This is also your semi-regular reminder that CellRanger is behind a laughably absurd and aggressive license, and obvioulsy works only on 10x chemistries.

lolz : clonesof.com

💥 You asked. RAW Lab delivered. DYNAMITE-ORFs v1.1.0 🧬 ✅ Prodigal gap controls ✅ Gene-calling fixes ⚡ E. coli auto mode: 20.6 → 9.2s in our benchmark Messy genomes. Meet DYNAMITE. crates.io/crates/dynam... #Bioinformatics #Rust

This looks very cool! github.com/Tussl-Co/Tus...