Nearly 7,000 LCMS runs to assemble a pan-cancer proteomic map of 949 cell lines!

It's here! Real LCMS proteomics at a biologically relevant scale. It takes a lot of time to generate something like this, obviously (almost 7,000 DIA LCMS runs!) and this is just the beginning, but this is EXACTLY what we need to prove that LCMS proteomics is here and can do real, relevant, biological studies. 

Oh yeah! Here is the paper. It's a preprint, with all the normal pre-requisites for skepticism, and I should absolutely not be offered a chance to review this because I'm thinking I should look up all of these authors so I know what they look like so they don't pay for drinks in bars if I'm in them. 

I need to go do stuff but here's how they did it: 

90 minute LCMS gradients with 2ug of sample.

5uL/min flow rates! 

SCIEX 7600 running variableSWATH/DIA. 

A bunch of technical replicates

Spiked in internal standards in every injection.

Think the mass spectrometrists are good? (Me too!) The bioinformatics is just as legit.

Spectral libraries were made with DIA-NN using the library data and it looks like that was the main processing tool...then it goes right into awesome land from there. 

My one and only criticism of this amazing and inspirational study is that they don't have a beautiful and well-executed data portal linking this proteomics data to historically accumulated data on these cancer cells. 

Just kidding. They also did that! You can check out Cell Model Passports here!
 

Quick math: 6,300 injections at 90 minutes ~394 days of acquisition time!