Thread by @DavidMorgenst17

Our paper is out!
Optimized Glycopeptide Enrichment Method–It Is All about the Sauce pubs.acs.org/doi/10.1021/ac…
#glycotime #TeamMassSpec #proteomics
A personal note (and couple of comments about publishing methods): \start

I really like this project, mainly because what it represents for me. I'm a scientist at a proteomics service facility led by @YishaiLevin. We're a bunch of talented scientists (if I might say so..), and our job,

in my opinion, is to enable excellent science through excellent proteomics. Enabling science is not only performing the experiments, but also making sure our capabilities are top-notch. This means investing in R&D of methods, and this is how this project was born:

While performing glycoproteomics analyses for a collaborating lab, we became dissatisfied with the performance of current methods. HILIC, multi-lectin and boronic all gave a couple of hundred quantifiable glycopeptides, and that was simply not enough.

We attempted to improve available protocols and capabilities, and improving Boronic acid enrichment seemed to be the most promising.
So this is what enabling science means to me - making sure that the tools we use for biology are as sharp as possible.

Another thing I like about this paper is that it takes a radical turn away from the "Numbers Game" so common. Lots of methods papers tend to show very high ID numbers, which are meaningless in practice: if you need to spend 10, 20 or 50 hours of instrument time per sample,

Then there is no chance I will be able to use that method in an actual experiment with 100, 500 or 1000 samples. What we're actually thinking about, is whether to continue running our 3hr gradients or reduce them to 90 minutes.

The over emphasis on ID's also makes quantification, which is the end goal most often, disappointing. in our paper ALL the numbers are of quantified peptides. ID numbers are slightly more than double. In a project we're running now, we've ID'd >13000 glycopeptiforms,

but quantified (with replicates) only ~7500. throwing around the higher number is impressive, but not useful.
And I think reviewers should insist on this point - if the proposed method is impractical, why allow it?

Lastly, couple of comments:

1. the quant numbers in the manuscript are conservative. as our Tech became more comfortable with the protocol, these numbers grew by ~30-50% in many projects.
2. switching to stepped HCD from EThcD really pushed the numbers up. we're getting 6000-8000 PSMs,

and 4000-5000 ID'd glycopeptiforms per sample. we usually lose 30-60% of them when we look at quantification. the 7500 quantified glycopeptiforms came from a set of 28 brain samples, using 90min gradients on an HFX instrument.

So, in conclusion, we're pretty happy with the outcome. we think our protocol optimization will, after a long time, make boronic acid a more popular method for glycoproteomics and will (hopefully) increase the depth of the observable glycoproteome.
And now, back to work 😉 \end.

@threadreaderapp unroll