Luke Gamon

Ah bugger sounds annoying. Especially for a quick ITI clean.

I’m starting to get confused. Is captivespray2 actually an improvement? Seems odd that they wanted to improve user friendliness and then made it easier to break the glass capillary.

Too-tight ITI sounds interesting, haven’t seen CS2 hands on myself, is it an issue specific to the new source?

Peer review is wild. Totally wild. Mostly very constructive and valuable criticisms, discussion points, feedback, details for improvement…. And then just a straight dig that seems completely unnecessary?

Depends how biased the AI review gets at reviewing AI proposals

Yup. If you use Positron via the databot you need to purchase tokens and run via the API. From what it’s quite a bit cheaper finding a way to use it via a subscription though.

TBH unclear yet whether it’s ‘worth it’ versus the time investment of validation. You know as well as I people are going to run with this though, with or without proper guardrails. For me right now it’s actually more of a coding assistant. No question it’s a massive booster in that scenario.

Ha!!! You’ve given me to perfect title for my upcoming pooblication.

One would hope that a carefully crafted proposal would beat out the vast majority of slop but I absolutely see your point about dropping success rates and incentives. Isn’t this just accelerating us towards something we always knew we would have to reckon with? Incentives in academia are a shambles

Definitely a learning curve but has big potential if you understand your data structure well and have a good idea of ground truth etc

Burned through $4USD last night in 2hrs of Claude-powered data exploration with DataBot in Positron. Absolutely addictive. Creates insights faster than you can validate them.

Fully reproduced the major hits from a multi-modal data integration, with plenty of new ones to explore/validate.

Nice! Sounds very useful as a coding assistant. Have you played at all with Databot for exploration?

Started playing a little with DataBot and Claude. Any tips and tricks? Are you just using chat or also for agentic coding with DataBot? Trying to wrap my head around tokens and context window.

GitHub - EuBIC/EuBIC2027: Organizational repository for the EuBIC-MS Hackathon 2027 Organizational repository for the EuBIC-MS Hackathon 2027 - EuBIC/EuBIC2027

📢 Mark your calendars!
We’re excited to announce the EuBIC-MS 2027 Hackathon!

📍 Gimo, Sweden
📅 31 January – 5 February 2027

💡 Have a hackathon idea?
The call for topic proposals is now open on the EuBIC GitHub repository: github.com/EuBIC/EuBIC2...

We’re looking forward to seeing you there!

We are in good company for sure

No wait found it again

Uh oh… TimsTOF disappeared

Hot damn I need to try this. Will it do image recognition on ggplot outputs etc if you tell it to output as a png or pdf etc?

Thanks! Great tips. Are you running R directly through CC? I assume you prompt it to run analyses, print the outputs or export plots, which it can then consider whether they look right or not, then iterate? I’ve been using LLMs long enough to know they’re not perfect but keen to see what can be done

Oh cool!!! Looks like you are limited to Positron supported models though?

Is there anyone out there who has experience implementing agentic code AIs with R/RStudio? I’m keen to try out eg Claude Code and I’m a little sick of copy-pasting code snippets between ChatGPT/Claude and R.

Is it just because of the extra mobility dimension? It is a serious pain. We are used to it since we only have Bruker instruments but damn it’s crazy fast when I download .raw from PRIDE to play around with.

It is my great honor to announce that registration for the 7th ESCP Single Cell Proteomics Conference is now open: lnkd.in/e4iyiQjf
With around 250 participants, it is one of the largest SCP conferences worldwide. We are also proud to announce that there is no participation fee for our conference.

I haven't automated it yet but for larger studies I usually do an ultra-fast DIA-NN search with a small spectral library as the samples are running basically live on the instrument. As you say - file size is a great indicator for major issues. I like the DIA-NN output for monitoring ion mobility.

DIANN for the search, custom R scripts and shiny app for visualising. Also relying a lot on the DIANN pdf outputs.

Also extra obnoxiousness would force me to point out that this is an aeropress not a french press. Lucky for them I’m stuck home with the kids.

Christmas party game trolling someone who ‘is a fancy Nisse, not satisfied with just water or cheap coffee, not like you, peasants!’

HA! I’m getting trolled at the floor Christmas party.

TBH I’d be shocked if everyone didn’t guess this one though. I’m clearly the only person obnoxious enough to walk up and down the corridor hand-grinding coffee beans.

It is also fastelavnsboller season 🤤

We’re hiring – Computer Science PhD at iBIO (Helmholtz Munich / LMU).

Build AI-driven, whole-body multi-omics atlases by combining scRNA-seq, spatial transcriptomics, proteomics & 3D imaging.

Looking for strong omics data analysis skills with ML/DL.
Apply or share:
www.linkedin.com/jobs/view/43...

I’ll be teaching this tomorrow in the context of paracetamol/NAPQI detoxification. One of the rare times I get to reference my own work.

Well, I teach on one of the biochemistry courses for our med students and can say that I use the word Protein as many times as humanly possible. I could easily go virtually the entire semester just saying ‘enzyme’ though.