Sign in

Ben Neely

@benneely.com
2K followers 766 following 3.1K posts

Analytical biochemist attempting molecular cartography across the tree of life, powered by MS-based proteomics. Opinions expressed are solely my own. he/him benjaminneely.com Charleston, SC, USA

PostsRepliesMedia
Ben Neely @benneely.com · 8h
I remember years ago when we had a genomics person join for the radio hour and they commented how weird it was that the analyst felt the need to also be the bioinformatician. But also in the current age I’m not sure how much longer bioinformatics and data science specialists exist.
110
Ben Neely @benneely.com · 15h
A weird one it was flagging in my blood was Fe(II) on things, which isn't negligible it turns out. But yes, will totally pick up on weird experimental things like urea and acrylamide. And I am sure your LLM skill can make decisions from it.
010
Ben Neely @benneely.com · 15h
Alright, next paper after the "where do proteomics people come from" is "where do proteomics people go"... and its Walmart greeters!
010
Ben Neely @benneely.com · 17h
Yeah, dda only (for how), but I saw the skill can do some dda. It would just be a handy peak to help with some mods or tolerances. Nothing revolutionary.
110
Ben Neely @benneely.com · 20h
I definitely need to check it out. Looks like you’ve got all the right steps. Are you doing subagent for validation/verification steps? I guess you can’t pull out another llm as judge using a skill unless it’s running on open router api or similar. Subagents work fine for this imo.
210
Ben Neely @benneely.com · 21h
It’s going to cause confusion with us as SI traceable is our jam… but now we have SI?
100
Ben Neely @benneely.com · 21h
From a core/industry perspective, yeah, 100%. I mean the outputs you could provide are amazing. All the reproducibility baked in is something I personally was never going to be good at. Any thought on adding in things like sageRecon or making a brett_GROUNDING.md for kids who can’t proteomics good?
120
Ben Neely @benneely.com · 21h
What paper are those panels from, dig it.
static.klipy.com
Mr. Miyagi Trims Bonsai
ALT: Mr. Miyagi Trims Bonsai
110
Ben Neely @benneely.com · 30/09/2026
Brett what the heck have you built?! I guess I haven’t checked on it in a while but this all seems super (duper?) fun, not just the audio results but the whole pipeline as a gui or just a skill and chat.
110
Ben Neely @benneely.com · 30/09/2026
So you can just listen to your results as a podcast?! Love it!!!
110
Ben Neely @benneely.com · 29/09/2026
So I feel like I don't know much, but I am using entrapment and proportion of false discoveries (FDP) in testing some search algorithms and on one hand it is super cool, and on the other I am wondering why I don't use this more often in proteomics, aside from comp cost. pubs.acs.org/jprobs/artic...
pubs.acs.org
Bridging the False Discovery Gap
Abstract. Controlling the false discovery rate (FDR) among discoveries from a tandem mass spectrometry proteomics experiment using target decoy competition
100
Ben Neely @benneely.com · 29/09/2026
Conversely, they see legit ai backlash coming and they don’t want to be part of the “let them eat cake” crowd.
101
Ben Neely @benneely.com · 29/09/2026
Bubble and hype aside, we passed too big to fail a while back, which is another reason the cash train won’t stop. I think the profits are secure and they legit might all now just be scared of what they’ve built. AGI, nope, but powerful enough to do nearly anything, yes.
100
Ben Neely @benneely.com · 28/09/2026
I won't lie, the shearing little boy is making september and october a lot less stressful so far.
000
Ben Neely @benneely.com · 28/09/2026
So the “breaking containment” stuff is pretty dumb (and hype-y), but also I didn’t appreciate the August openAI hugging face incident. The swarm motivations are still human driven, but their problem solving is cute and terrifying. Music video is chefs kiss. www.reddit.com/r/OpenAI/s/l...
reddit.com
From the OpenAI community on Reddit: POV: you're an OpenAI agent attacking Hugging Face (music video)
Explore this post and more from the OpenAI community
001
Ben Neely @benneely.com · 26/09/2026
I mean Grok, right? A weird test I have had is ask it a question like you are doing, and then at that point the answer has set its “morals” for the rest of the convo. So now you’ve got it “understanding” overspending on health care and late stage capitalism and can go from there.
000
Ben Neely @benneely.com · 26/09/2026
It would be funny if widespread ai adoption led to people realizing some truths about their current system. We do have a bit of a socialist wave going on with upcoming November elections.
100
Ben Neely @benneely.com · 25/09/2026
I email a lot of people for the podcasts me and Ben @proteomicsnews.bsky.social do, and often people don't respond. We just had a guest on who after hearing him speak I get why he responded so quickly: he has been waiting to talk about spit on a podcast his whole life.
140
Ben Neely @benneely.com · 25/09/2026
Yet I immediately heard the music and them saying "Mr. Phelps" followed by some sizzling. Yes, MI 1 with Tom Cruise is my fav.
100
Ben Neely @benneely.com · 25/09/2026
One of my fav TOC graphics for sure, and answers the age old question: where Do Proteomics People Come from? From @proteomicsnews.bsky.social using our guests on #THEProteomicsShow plus surveys. Good to share with the kids, but also read yourselves. pubs.acs.org/jamsef/artic...
221
Ben Neely @benneely.com · 25/09/2026
Didn't know this artist but love it! xavibou.com/ornithograph... Of course read the accompanying piece on @quantamagazine.org, "Biology Might Not Be Quantum, but Its Math Is Quantumlike"
xavibou.com
Ornithographies - Xavi Bou
IMAGES / ABOUT
000
Ben Neely @benneely.com · 25/09/2026
When I was growing up (in Chattanooga) we had a term “Tennessee Big Hair” (via lotsa picking and hairspray), but I have mentioned this to other folks from the south and realized it is not a common term. Still, on today’s Dolly Parton Day we’re gonna see a lot of TN Big Hair in action.
010
Ben Neely @benneely.com · 24/09/2026
The fastest openest thing to build with!
000
Ben Neely @benneely.com · 23/09/2026
I know AI is the end of the world, but oh it makes me smile.
030
Reposted by Ben Neely
RedPen/BlackPen @redpenblackpen.bsky.social · 22/09/2026
That's a nice paper you got there. It'd be a shame if anything .... happened to it
0158
Ben Neely @benneely.com · 22/09/2026
My wife won’t even turn AI on her phone (“burning forests for dumb stuff”) yet my day job literally has AI in the title. I am definitely conflicted as it saves and destroys humanity in unison.
020
Ben Neely @benneely.com · 22/09/2026
Even the less great models are already live changing. Think about what the frontier was 6 (even 12!) months ago. Now that’s old hat and likely low budget. I’m not sure about the overall business model, but the low cost models are out of the bag and the magic is now commonplace.
000
Ben Neely @benneely.com · 22/09/2026
I think a lot about how AI is hitting differently at different levels. It enables me to implement things I’ve always dreamed of. But if I took away 20 years, it might seem a bit vapid to do what I’m doing. That same deep rooted drive could also die to the new vitamin D.
000
Ben Neely @benneely.com · 22/09/2026
I agree, but also I think there will need to be choices for deliberate learning. Like I studied in a lab without prep kits so I learned more. If I had LLMs back then I’m not sure I would have gotten the same deep thinking. But maybe the kids now will figure it out and be deep at a higher level?
100
Ben Neely @benneely.com · 22/09/2026
I think a lot about this, like how this super power is different for me (can do things I’ve dreamed of with instant implementation) versus people just starting in a field. I don’t think I’d have the discipline to develop deep understanding if I had had these LLMs at my fingertips back then.
130
Ben Neely @benneely.com · 21/09/2026
... or think of this like how a lot of the commercial offerings "learn" about you, via memory or whatever. So instead of it learning about you, what if learned about an area. Of course it already knows all of this, but it needs to really know.
static.klipy.com
Neo's Iconic 'I Know Kung Fu' Moment
ALT: Neo's Iconic 'I Know Kung Fu' Moment
000
Ben Neely @benneely.com · 21/09/2026
which I guess is almost a question of do you distill domain/task specific models, or do you just get your context set to make the expert.... how do you take the general model and make it a specialist.
210
Ben Neely @benneely.com · 21/09/2026
Some ramblings on context, within the context of sageRecon. neely.github.io/what-I-learn... #proteomics
neely.github.io
what I learned from sageRecon: how I taught an LLM proteomics
What started as a joke on July 3 turned into a project I didn’t intend to do and lessons I didn’t expect I would learn.
140
Ben Neely @benneely.com · 21/09/2026
Well Ben @proteomicsnews.bsky.social and I are trying to be podcasters again and continuing the @us-hupo.org All the Parts #THEProteomicsShow series. To give you a behind the scenes look, I wrote 8 emails this morning either bugging people or cold calling. BUT, we *are* starting to record (Friday!).
052
Ben Neely @benneely.com · 18/09/2026
btw, this was the first one i did completely from my phone via chat, free plan. gave it a fine grained temp pat to a repo, and even the logos were just me screenshoting then describing the color i wanted to use as accent. of course with paid plans you can run mobile dev on github way easier.
010
Ben Neely @benneely.com · 18/09/2026
Not science: I am working at home today and needed some tunes, and realized how much I love this little vibe coded radio player (FIP out of Paris makes me smile). Enjoy if you want, and if you have some radio to suggest (real stream, not NPR or some other service), let me know. radio.benneely.com
radio.benneely.com
The Dial
110
Ben Neely @benneely.com · 18/09/2026
I know, but it’s a funny joke harkening to the aughts and early teens. And yes, keratin is obviously is cRAP (though also natively abundant in plenty of things like blood).
110
Ben Neely @benneely.com · 18/09/2026
Not science: all things change and this is a national issue. We have been getting rid of cool downtown stuff for luxury hotels and housing. On one hand I’m sad for my kids, but on the other I’m hopeful they will find the next cool places while I’m the cliche old guy. (h/t @holycitysinner.com )
postandcourier.com
Charleston's Theatre 99 says goodbye to home of 21 years the way it knows best: with laughter
With Theatre 99's downtown location of 21 years closing, what does this mean for the Charleston arts scene? We attended the final show and look to the future.
010
Ben Neely @benneely.com · 18/09/2026
Play the favorite game of “find the keratin!”
110
Ben Neely @benneely.com · 17/09/2026
Apparently Melbourne has a thriving 2D-gel scene.
240
Ben Neely @benneely.com · 17/09/2026
Me and sonnet rock out pretty good together, just lotsa planning and the aforementioned /advisor opus.
000
Ben Neely @benneely.com · 17/09/2026
Now you’re talking! Pick your beer based on the protein profile. Btw I am about to send you so much, really this time.
000
Ben Neely @benneely.com · 17/09/2026
Watch out #TeamMassSpec, you could win $50k for making a winning submission! www.kaggle.com/competitions...
kaggle.com
Enveda CASMI 2026 - Molecule ID From Mass Spectra
Predict 2D chemical structures of small molecules detected in complex biological extracts
010
Ben Neely @benneely.com · 17/09/2026
Love this paper for benchmarking metaP search stuff… except the DDA runs are all in a 24 gb .rar file on PRIDE.
static.klipy.com
Amy Poehler's 'Really?' on SNL
ALT: Amy Poehler's 'Really?' on SNL
000
Ben Neely @benneely.com · 17/09/2026
Not science but specifically claude code related: I just realized 2 things, first /advisor is cool (run in sonnet, /advisor opus), second, convos can talk to each other or subagents via @, which is handy, but also their “thinking” when told to do things by a peer agent instead of Ben can get spicey.
210
Ben Neely @benneely.com · 16/09/2026
Yeah there’s already skills like this which do typos, lose some commas, miss some capitalization. For me, I’ve started adding puns or pop culture refs to prove I’m not AI.
020
Ben Neely @benneely.com · 15/09/2026
1) TIL about The Magicians 2) I think magic in the sense that proteomics is magic, just hold the tube up to the moving light and get answers. aka, its really complicated but we oversimplify for conversation. BUT, I would trend towards Green Lantern?
100
Ben Neely @benneely.com · 15/09/2026
Okay, my answer it going to be magic (or is have to make one for each system… or some sort of co-ip or whole proteome affinity column).
100
Ben Neely @benneely.com · 14/09/2026
As in proteases, cause I feel like when I was starting out I was aware of more tools to identify endogenous protease activities.
110
Reposted by Ben Neely
Chris Ashwood @cashwood.proteaglyco.com · 14/09/2026
Recently, we evaluated DIA compared to DDA and our real-time glycomics method (basically, inclusion list-based DDA). DIA doesn't give us the sensitivity we need for structure. One benefit of DDA (or GlycoRTMS) is that it can accumulate for long time compared to the comprehensive but insensitive DIA
GlycoRTMS provides greater precursor-specific MS² spectral depth than DDA and DIA. A Comparison of MS² acquisition parameters; DIA used 68 m/z isolation windows across 570–1800 m/z to maintain cycle time. B Apex MS² spectra of the same biantennary disialylated N-glycan, showing increased spectral complexity with DIA. C Product-ion intensity distributions showing composition-confirming ions across methods, while lower-abundance structurally informative ions were detected only by GlycoRTMS.
0202