I don't know who needs to hear this, but if you want keywords from proteins that recently got pruned from UniProtKB, you can use Unisave. I wrote this post which has an associated github repo to help. OR just switch over to 2026_02 ASAP. They told us it was coming. neely.github.io/recovering-u...

Recovering UniProt Keywords for deleted accessions

If you searched against a pre-June 2026 UniProt database, some of your identified proteins no longer exist in the current release — and that means no keywords.

neely.github.io

I keep thinking back to @aptshadow.bsky.social’s crows and how “load up the model’s context window in a way that makes it more likely to predict, or ‘approximately retrieve,’ reasoning-shaped strings of text” is just a fast march to “jagged intelligence” leading to 42 (or maybe it’ll be furty too).

Quanta Magazine@quantamagazine.org · 4d ago

What are AI systems actually doing when we say they’re busy “reasoning”? In this week’s Qualia column, John Pavlus reckons with the intellectual whiplash of the last few years. https://www.quantamagazine.org/is-ai-reasoning-right-for-the-wrong-reasons-20260731/

Having AI makes everyone their own CTO/idea person, right. I’m on vacay and can tony stark some ideas/tasks to my assistant and they knock them out and report back, all while I’m drinking one coffee. Imagine an autonomous lab that never sleeps that I could feed experimental questions/tasks to! 🤯

I am zero % surprised on this. That paper we just had out in JPR had a handful of pop culture refs I normally wouldn’t leave in, but was me screaming “I’m a human!” Also, I was in a student presentation last week that had freestyle drawn parts in figures… lazy or making it human?

WIRED@wired.com · last wk.

Novelists, journalists, and power LinkedIn posters are embracing first-person narratives and idiosyncrasies to avoid being mistaken for chat bots. www.wired.com/story/more-t...

I get to take off for a few days, and thought yesterday I was wrapping up a project. Nope, not quite back to square one but quite a few steps back. Me and metaproteomic analysis, is fun until I realizes I wasn't getting search results for all the taxon I asked for. neely.github.io/It-is-going-...

It’s been a blast seeing how the proteomics field has developed over the last 26 years – I hope that great things continue to evolve over the coming years as well. Signing off – and keep well everyone. 3/3

RNA abundance positively correlates to protein abundance.

Richard Sever@richardsever.bsky.social · 2w ago

The Impact Factor is an excellent measure of article quality.

Not science: one of my fav things about Pogi is the concept of him doing side quests. Today is one of those: does he try and break Pantani’s record or help Del Toro? My fav side quest: last monument and classics. Also, @pwilmarth.bsky.social would point out the women start today.

Trigger warning, DNA! I bought this shirt to wear at proteomics things. Could be fun to get the DC RF shirt too that I think I saw @lindsaykpino.com with. And yes, woot shirt is running a buy 2 get 2 deal.

HPRC2: A human pangenome reference with near-complete coverage of common genetic variation https://www.biorxiv.org/content/10.64898/2026.07.21.739710v1

Still my fav thing to do or think about for proteomics: forensics. A lovely little collection in JPR, will come back and try and tag the authors here, but also h/t @magnuspalmblad.bsky.social for sharing. pubs.acs.org/doi/10.1021/...

Collection on Forensic Proteomics

This publication is free to access through this site. Learn More

pubs.acs.org

Self-promotion or PSA, a little of both? Our (@magnuspalmblad.bsky.social and Jared Ragland) Letter is out in JPR, which tackled the age old question "how can we make AIs follow community-supported domain knowledge?" Answer: use a GROUNDING.md file in your context. 1/7 pubs.acs.org/doi/10.1021/...

Agentic AI-Assisted Coding Offers a Unique Opportunity to Instill Epistemic Grounding during Software Development

The capabilities of AI-assisted coding are progressing at a breakneck speed. Chat-based vibe coding has evolved into fully fledged AI-assisted, agentic software development using agent scaffolds, wher...

pubs.acs.org

The update zero people wanted: I am way too distracted by this. I have learned too much about certain steps, and incredibly thankful that @michaellazear.bsky.social's Sage is stupidly fast. I can do a series of searches in under 3 min (high end), which makes development and testing a breeze.

Ben Neely@benneely.com · 4w ago

AI trigger warning What I’ve learned so far: 1) scope creep is really hard to fight 2) PTM Shepherd was clever 3) there isn’t an obvious exhaustive ptm dataset to validate on, more of a DIY or algo competition approach for vetting 4) I like Cline 5) I have other things I’m supposed to be working on

Surely someone is gonna be making ice man jokes with the media frenzy that’s ensuing? But hey, at least we are one shotting macOS instead of Minecraft or doom?

Kimi Raikkonen 'Deal With It' F1 Meme

ALT: Kimi Raikkonen 'Deal With It' F1 Meme

static.klipy.com

I think about this a lot. Like sure, these tools are great, but not “kill society and world” great. If they don’t pop out some huge discovery, cure or advance pretty soon, I think they’ve likely overstayed their welcome… though now it’s an arms race and we know how those end.

Marie Vibbert@reasie.bsky.social · 3w ago

No science fiction could have prepared us for just how stupid the end of the world can be. "The environment was on the brink of collapse, and so, rather than deal with that, we burned what fuel we had left getting computers to scam each other for money."

Do you want to work at the frontier of structural and chemical #proteomics? Interested in induced proximity and building a feedback loop for machine learning models to explore biology? I'm hiring! Apply below ⬇️ jobs.ashbyhq.com/proxima/6e39...

Proteomics Scientist

Proteomics Scientist

jobs.ashbyhq.com

ai trigger warning I really don’t mean/want to post lotsa ai stuff, but this repo template is making me super happy. It’s made for cold start (with a defined shutdown protocol), so you can come back whenever with whatever LLM (give it r/w on your repo). Directions within. github.com/neely/agent-...

GitHub - neely/agent-context-project-template: A template for a repo-level context system for solo development on main, built to survive a cold start — a fresh chat with zero memory should be able to ...

A template for a repo-level context system for solo development on main, built to survive a cold start — a fresh chat with zero memory should be able to read these files and pick up exactly where y...

github.com

🐐 How do we decide what to sequence next? Genome on a Tree (GoaT) brings together taxonomy, phylogeny, sequencing progress, and sample availability to help identify the species that will have the greatest scientific impact. Has your favorite species been sequenced? 🔍 goat.genomehubs.org #OpenScience

Qualia goes where curiosity leads. This week, columnist Natalie Wolchover tracks the secret of left-handedness through developmental biology, genetics, evolution, and culture. www.quantamagazine.org/why-am-i-lef...

Why Am I Left-Handed? | Quanta Magazine

An invisible difference in 10% of humans poses deep mysteries in several fields at once.

quantamagazine.org

Excited to share our new review in Current Opinion in Chemical Biology! We discuss how advances in proteomics are transforming our understanding of where, when, and how signaling occurs in cells. Thanks to @alexmatia.bsky.social for writing the article with me. www.sciencedirect.com/science/arti...

Resolving cellular signaling in space and time: From organelle proteomics to spatial phosphoproteomics

Cellular signaling is inherently organized in space and time, requiring coordinated control of protein localization, molecular interactions, and enzym…

sciencedirect.com

Not science/TMI: I made this post about it being hot but having AC. Last night my HVAC in the house tried to break, but I think I have it working (after an hour of sweltering in attic), and today on way to work the fan belt on wife's car shredded, killing the AC among other things. Laws of 3 coming!

Ben Neely@benneely.com · 4w ago

Not science: I think the thing people may not appreciate about certain heat, isn’t how hot it is during the day, but the fact it never cools off. Thank the tiny gods for AC at night. Also, it being hot in places that aren’t used to heat or don’t have AC (or southern France and Le Tour)… ugh.

Evotip flow through for gDNA. Nice! But made me wonder about new hotness in other methods: what happened to nano pots where they put a tissue on the spots to do multi omics, or floating cells from @gaolab.bsky.social? I’m not in this field and this is a lazy post. (👀 @proteomicsnews.bsky.social)

Mann Lab@mannlab.bsky.social · 4w ago

New preprint: Cell-type-resolved spatial proteogenomics, matched genome AND proteome of the same cells. The trick: the flowthrough we normally discard after loading peptides onto an Evotip contains genomic DNA at 85–95% yield. One tip, two molecular layers. www.biorxiv.org/content/10.6...

Not science: I think the thing people may not appreciate about certain heat, isn’t how hot it is during the day, but the fact it never cools off. Thank the tiny gods for AC at night. Also, it being hot in places that aren’t used to heat or don’t have AC (or southern France and Le Tour)… ugh.

AI trigger warning What I’ve learned so far: 1) scope creep is really hard to fight 2) PTM Shepherd was clever 3) there isn’t an obvious exhaustive ptm dataset to validate on, more of a DIY or algo competition approach for vetting 4) I like Cline 5) I have other things I’m supposed to be working on

Ben Neely@benneely.com · last mo.

Alright proteomics skywalkers, a challenge: we all love open search/error tolerant search/delta mass/Preview… so let’s race to build a Sage-based algo to predict ptm dist. plus ragged ends, enzyme efficiency, etc. Added bonus if you know a good benchmark data set. Then put in JPR tools issue. Go!!!

LLM/vibe trigger warning It’s not efficient with tokens to have the LLM handling git to some repo but it makes it way easier. I was manually updating files but now I can have it capture all the iterations. Even in a “big” project giving it a PAT and writing to a dev or feature branch would be fine.

Ben Neely@benneely.com · last mo.

I think I’m done it, at least the main first iteration. github.com/neely/radio

Not science: the @scalzi.com collection on @humblebundle.com. I really do enjoy Scalzi from Old Man’s war to Collapsing Empire to whatever else. I’ve read nearly all of this, and the ones I’m missing I’m doing audio (redshirts).

DIA proteomics needs more than powerful search — it needs transparent, reproducible, and scalable software. I’m excited to share that Radiant DIA™ and the Fulcrum Scalable Pipeline™ are now released with publicly available source code.

Took me a while to realize our local 250th events were somewhat unique… we’re a signature city! Fun part is that meant we had some new extra big fireworks in the harbor tonight. americasnationalparks.org/americas-250...

Alright proteomics skywalkers, a challenge: we all love open search/error tolerant search/delta mass/Preview… so let’s race to build a Sage-based algo to predict ptm dist. plus ragged ends, enzyme efficiency, etc. Added bonus if you know a good benchmark data set. Then put in JPR tools issue. Go!!!

I was about to quote my fav Arlo Guthrie song, but somehow this is autocorrect. youtu.be/BvLtNBm1yyA?...

Arlo Guthrie - Motorcycle Song

YouTube video by PatchouliHaze

youtu.be

PRIDE Database@pride-ebi.bsky.social · last mo.

PXD057854 🚨 PICKLE1-3 isoforms of human PRICKLE1 and PRrickLE2 are a key component of the Wnt/Planar cell polarity signalling pathway. 🚨 New dataset alert! 🚨