2,000 plasma proteomics samples in 6 weeks AND pQTLs by LCMS!

 

Yesterday was the last day of the NBA regular season with loads of playoff implications for all teams with places 5-9 in the western conference decided in 1 day! 

Stick with me here, because the proof that proteomics is BIG DATA and can do BIG COHORTS are starting to drop and I think there is similarity here. 

We probably can't stop the "next gen" proteomics tools that are coming. They have too much inertia and too much money, and even when there isn't a lot of proof they work yet, we'll either have it soon -- or we won't. 

In the meantime, they're resetting the expectations for our field. Worst case scenario -- it turns out that things like O-link and Somascan and WhateverIt'sCalled suck and we forget about them 5 years from now -- they've done one major thing.

They've proven, without a doubt, that medical researchers and biologists all want big cohort proteomics data. They appreciate the value, they just don't think LCMS can do it. 

Now, you could argue that worst-case scenario is that these new technologies are amazing and they displace us, but the season ain't over yet. Want proof?

>2,000 plasma proteome samples. It took some work to find out the throughput, but based on the references and the fact this paper used EvoSep 60, I think that is what it is, so not counting controls and libraries, that's 1 month for one Exploris (DIA). 

That's great and everything, right? This isn't the coolest part for me. It's the pQTLs. 

Quantitative Trait Loci (QTLs) are a big deal for bioinformaticians. There are entire fields of researchers looking at the genome/transcriptome/proteome in a different way than we are. They aren't just isolating this one thing that we think is a distinct protein and trying to figure out if it is linked to things. QTLs focus on phenotypes -- in humans, that's going to be the clinical information -- and then look at what quantitative changes exist when that phenotype is present. It sounds like what we do, but the subtleties and the direction at which you look at it are very different.

This probably explains it better. 

On a more granular level -- the people interested in using O-link are probably interested in proteinQTL output. (pQTL). 

What did this paper do? It used LCMS to produce pQTL output, in addition to what you and I are more used to seeing. 

Back to the last day of the season -- It's like Matthias Mann wanted to do whatever the f*(&* Jonathan Kuminga did to this rim in Q3 yesterday. 

And we've still got the playoffs!