Summer Intern, Data Science - Seattle, WA Note: To apply for this internship, you MUST submit both a cover letter and a resume, both in a single PDF (Word and .doc files will not be accepted)! Talus Bioscience is seeking highly motivated sum...

🧬 We're looking for an intern for summer 2025 to work with me and the data science and engineering team at Talus Bio!

Apply here 👇
talusbio.applicantpro.com/jobs/3551253

Something else for me to worry about. Thank up for that 😅

Decided to take Amtrak down to Portland for US HUPO. I give myself 50/50 odds of being on time lol

A learned score function improves the power of mass spectrometry database search bioRxiv - the preprint server for biology, operated by Cold Spring Harbor Laboratory, a research and educational institution

Lots of people use machine learning to post process mass spectrometry database search results. But why not just use ML as the score function in database search? Turns out it works great! www.biorxiv.org/content/10.1...

I'm compiling a page of proteomics references that people forget or put up incorrectly. ProteomeXchange, FragPipe, SP3, S-Trap are all ones I see left out or wrong (or not what the resource owners asks you to use) are there others?

I have always personally thought larger the better (eg 1:100).

The dynamic programming part of MS-GF+ would probably be a fun (re headache inducing) challenge for undergrads

New faculty position available for computational biology in the department of Genome Sciences at the University of Washington. Our department spans both Genomics and Proteomics. apply.interfolio.com/135108

I should note that one needs to remember the existence of "neighbor peptides" if you just search on the subset. Neighbor peptides are irrelevant peptides that look like relevant peptides.
pubs.acs.org/doi/10.1021/...

I agree that different solutions are required for different use cases. I would even argue that for certain cases (really small database) that FDR is the wrong thing to do.

I think it was 2017 @neely.bsky.social

Maybe I am missing something but I'm not aware of any DIA analysis that only looks for a fre peptides. Am I reading your ppst wrong?

Are you referring to the talk where he (and Uri) showed that different shuffling of the decoy databases can yield different estimates? This effect becomes larger as the db becomes smaller.

pubmed.ncbi.nlm.nih.gov/30560673/

I really like the wording of this approach. Going to have to remember it for the future.

Summary of my quant proteomics #Asilomar2023 talk:
1. sample space >>> throughput(e.g. chemical compounds in drug screening, cell types in biology, etc)
2. Interesting things are typically rare
3. Followup work will filter out false positives, but you can never recover a false negative

Is there a beef with mstdn? I've haven't seen anyone articulate that yet. I think we as a community are still trying to figure out the next step so to me it seems natural to have some chaos.

Themes I'm picking up on at Asilomar 2023 #TeamMassSpec conference 🧵

It's kind of interesting how much FDR has been talked so far. It's day 2 and there have been like 5 talks.

On the other hand, I don't remember the last time I saw a FDR focused talk at ASMS. Not sure what this means, if anything, but an interesting observation.

Which paper was that? I think I missed the thread mastadon.

Sorry not sorry, but another US HUPO 2024 related post: we are looking for nominations for the four annual awards (see link). The winner gets a lecture (among other things), so nominate someone you want to see speak. It's easy and fun, and will make Portland even more interesting!

Thanks! And I wil admit I typically don't do metabolomics either. Bit of a new experience for me.

We then looked at how MHNs may provide utility for annotation of metabolomics data and may aid in interpretability of metabolomics by applying this representation to several previously published datasets. (4/4)

In a MHN edges connect an arbitrary number of nodes. These edges allow for more complex visualizations of relationships that may be hidden in a graph representation. An example can be shown in Fig1, which describes how 3 different coauthor relationships (1B/1C/1D) would yield the same graph. (3/4)

A molecular hypernetwork (MHN) is an extension to a molecular network (MN) that has been popularized by platforms such as GNPS. A MN uses a graph representation where nodes are spectra and edges connect two nodes that have high spectral similarity. (2/4)

Alright time for a preprint announcement. In this manuscript we introduce the use of hypernetworks for the purpose of visualizing and analyzing metabolomics data. And before anyone asks, yes, this is the first time I have worked with metabolomics data before. (1/4)

*knocks on door*
Hey everyone, made it here. Time to figure out how this place works.