Lightnews — Scholar-powered news

Phil W @pwilmarth.bsky.social · 10h

US HUPO short course topic suggestions:

You are doing it wrong because no one knows how to do it correctly.

Techniques that actually work (*extremely* short course).

AI is to science like creationism is to evolution.

Experimental designs that tell you something: finding needles in haystacks.

2 2

Phil W @pwilmarth.bsky.social · 3d

Ha-ha. What are the project sizes? A few MB? Their combined size is probably modest. Get a 10 or 12 TB drive and archive all the old data on one drive, then put it in the desk, and you will have some top drawer data!

1

Phil W @pwilmarth.bsky.social · 3d

ChatStat or StatChat??? I bet this protein is always differentially something or another: www.uniprot.org/uniprotkb/Q5...

1

Phil W @pwilmarth.bsky.social · 4d

I gave a quick listen to Taylor Swift's new album. My first thought was that I hope they are not making a video for Wood...

1

Phil W @pwilmarth.bsky.social · 7d

There might be a download option to include canonical plus isoforms (however, canonical and isoforms are terms that only apply to the SwissProt entries).

1

Phil W @pwilmarth.bsky.social · 8d

See www.uniprot.org/help/uniprot... and www.uniprot.org/help/manual_... for more details. There is a lot of documentation at UniProt.

UniProt

UniProt is the world's leading high-quality, comprehensive and freely accessible resource of protein sequence and functional information.

www.uniprot.org

2

Phil W @pwilmarth.bsky.social · 8d

Isoforms are represented differently in SwissProt vs TrEMBL. Different protein forms have explicit, separate entries in TrEMBL. Those related sequences are rolled into one curated SwissProt record. There's one canonical, representative sequence and others are detailed in the record, not the FASTA.

1 1

Phil W @pwilmarth.bsky.social · 8d

Maybe the Government shutdown means the Federal Government won't be able to terrorize "we the people" of the US as much, or are those the "essential functions" now?

Phil W @pwilmarth.bsky.social · 11d

I bet the amount of coding by folks that know nothing about coding (and have no business coding) has gone way up since AI code breakers crawled out of the swamp.

1 3

Phil W @pwilmarth.bsky.social · 11d

There is nothing "civil" going on down here. Reporting from the war-torn Western Front...

1

Phil W @pwilmarth.bsky.social · 12d

The 'ome nameome is out of control!

1

Phil W @pwilmarth.bsky.social · 12d

Boy, I'm sure glad federal troops are coming to get crime under control here in Portland. On my walk today, I saw some cars not come to complete stops at stop signs. There were a few cars exceeding the speed limit. There were some pedestrians that jay walked. I saw some litter. It is out of control!

1

Phil W @pwilmarth.bsky.social · 12d

The "status quome" would be easy to study...

1 1

Phil W @pwilmarth.bsky.social · 13d

The beauty of DIA is that you make one dataset as a grad student, and you can just keep reanalyzing the same old data with each new software release until you retire! DIA really stands for data independent analysis...

2

Phil W @pwilmarth.bsky.social · 13d

Good FASTA file choices are more about quality over quantity (see digitalcollections.ohsu.edu/record/3031?...). You can't do protein alternative splicing or mutations or PTMs unless you have high protein sequence coverage. You need to see what high abundance proteins are in the sample, then refine.

The effects of parsimony logic and extended parsimony clustering on protein identification and quantification in shotgun proteomics

Madhira, Raviteja

digitalcollections.ohsu.edu

Phil W @pwilmarth.bsky.social · 14d

Good question. If you search for taxon number or species, you look at the proteome options in the left sidebar at UniProt. Click on reference proteomes, then look for download options. You are looking for the phrase "one protein per gene" in a download link. I always fumble around until I find it.

2

Phil W @pwilmarth.bsky.social · 14d

Canonical reference proteomes using the "one protein per gene" download option. You should have around 21K to 22K sequences for any higher eukaryotes. FASTA file choices are discussed a bit in this post: pwilmart.github.io/blog/2020/09...

Unique peptides and shotgun quantification

pwilmart.github.io

1 2

Phil W @pwilmarth.bsky.social · 15d

Like my computer skills, I would have to be a bike mechanic at a bicycle museum. I don't think I could even get a wheel off of a new carpet fiber disco brake rodeo bike. I could probably throw ebikes in the dust bin when they break (are they even serviceable?), but they are pretty hefty...

1 1

Phil W @pwilmarth.bsky.social · 15d

Cat herder.

1 1

Phil W @pwilmarth.bsky.social · 15d

Last generation sequencing?

2

Phil W @pwilmarth.bsky.social · 16d

It was to prove that infant vaccinations cause autism... Or maybe Johnson and Johnson did not sign up for a Trump Gold Card... Or the spray tan and comb-over daily prep gave someone a headache, they took a Tylenol, and that was the only medical term they could remember at the press conference...

1 1

Phil W @pwilmarth.bsky.social · 24d

If you have a system that is already too complicated to properly understand and test, is the answer really to make an even more complicated system? It is a self-destructive path for MS-based proteomics.

Phil W @pwilmarth.bsky.social · 25d

What is that in ROC speak? An AUC of 0.60? Ha-ha.

Phil W @pwilmarth.bsky.social · 25d

The tech industry solved the problem for me. I have zero relevant IT knowledge now. Unless a computer museum is looking for an IT person...

1 1 3

Phil W @pwilmarth.bsky.social · 26d

If you can't find a Far Side comic to make your point, you don't have a point to make.

1 3