Revisiting gas phase fractionation with today's technology!

Every few years someone (actually -- pretty commonly, one guy in Wisconsin) gets fed up with HPLC entirely and tries to do something smart so the pumps can finally be dropped off the roofs of the parking garages.

It always seems like the technology is just....almost....ready....

Here is one I found particularly promising 7 years ago....

We've come a long way since then! Are we there yet?  Maybe? Check this out! 

132 samples in 4.4 hours? Okaaaaaaay -- that's something.....

The name is...meh...for a method out of Wisconsin. Don't worry, Twitter, has it covered --

--what about.....

Meta-Analysis of 1,000 Proteomics Experiments in the Search for Missing Proteins!

This is a really interesting study describing the progress toward finding the "missing" human proteins (the ones that genomics says should be there, but we haven't been able to find protein data for).

--and reanalyzing a few (1,000!) of the files toward just one chromosome of the Chromosomal Centric Human Proteome Project (C-HPP).

An interesting take-away for me from this study is that it appears that we are approaching a state of diminishing returns due to how much effort goes into the identification of so few peptides. We've found the easy ones, I guess! Time to think up some new strategies?

Rereading -- FAST and Low-cost detection of SARS-CoV-2 peptides by LCMS!

After a few hours of sleep, let's looks at this awesome new preprint again!

What's it do? Well...it tries to establish our beloved LCMS systems and shotgun proteomics work into a real life diagnostic assay for the virus.

How'd they do it? They got some infected nasopharyngeal swabs and removed 200uL of the material for digestion.

They used SP3 digestion overnight (following reduction/alkylation) they prepped with a Hamilton Starlet --  I think it was this one, but this might be a newer model?

They used microflow at first on a QE HF-X. And that was fast -- that was like 9 minutes.

Then when everything looked great -- they went Turbulent Flow(??) with a Triple Quad (TSQ Altis) and cranked up the speed dramatically.

Do you know about these multiplex HPLC things? Here is an example image.

You see them in biopharma labs all the time. You run multiple gradients simultaneously and you either do the UV Absorbance or mass spectrometry on the point in each gradient that you're interested in.

For example, if your peptide of interest comes off at 33 minutes and that's all you care about, why waste that precious mass spectrometer looking at the whole gradient? Just switch over at 32 minutes, ionize till 34 and switch to the next column and gradient. HPLC pumps are $30k? Switching valves are $2k? 4 pumps and 2 switching valves and you're doing the work of 4 LCMS systems for less than the price of two.

Now...your plumbing flowchart kinda looks like this in the end ---

-- but as long as nothing goes wrong, you're getting high throughput on the cheap!

I'm not familiar with the difference between high speed multiplexing and Turbulent Flow, but I think this all still makes sense.

This is how they get to crazy throughput. In theory -- 500 individual samples screened per instrument per day. Which is fantastic!

Again. Great paper. I should sleep more. And read slower. Awesome work and a huge congratulations to this team.

It's glycoproteomics week! SugarQuant, MS-Fragger-Glyco and O-Pair search!

As a silver lining for this virus thing is that we've got lots of time to work on informatics pipelines! How much? What about enough to drop 2 new glycoproteomics workflows the same day?

The first is an extension of the ultra fast MSFragger workflow and the figure at the top does a great job of explaining what it is doing. 

The second (and in no particular order, when I select multiple images, blogger inserts them however it feels like) -- appears to be a new addition into MetaMorpheus, called O-PAIR search! 

I unabashedly love both of these software packages, but don't have the time to test them and write glowing things quite yet. 

On the glycoproteomics side of things I also have to point out this thing! 

SugarQuant is more than a software package -- it's a new twist on just doing glycoproteomics in general. TMT labeling + SPS MS3 for glycoproteomics and it utilizes ALL THE DATA. 

The software (GlycoBinder) can be found at this Github!

Whoa! Some guys came out of nowhere and got $100M to do proteomics!

Man....I've got the wrong friends....

Okay, I get it. The ProteoWizard guy does have a bunch of citations, but....ummm....almost $100M to do some vague proteomics stuff with some guy who definitely has not background in our field?
Good for them! If this turns into some silly array stuff, I'm going to vomit.

Check these articles out!

This one is even more interesting. Stealth mode proteomics.

I assume with all that money they'll be hiring a bunch of great people out there to do whatever it is that they are doing.

Amateur hour is over -- Mass Spectrometry Diagnostics for COVID-19!

EDIT: Better read through of this paper posted here.

When it is time to make a new clinical diagnostic. The ONLY time that mass spectrometry is considered.

The only time.

Is when there is literally no other option. No matter how bad that other option is.

People will bleed every drop of blood from every rabbit on earth before looking toward the most accurate analytical instrument ever developed on earth.

Well -- my friends -- in COVID-19 diagnostics, we've literally tried everything and we're out of options.  I guess it's time to go to mass spectrometry!

Wanna check it out? Or do you want to go suck the blood out of some rabbit and try and convince me that whatever that rabbit blood does is reproducible? If it's the former, here is the link!

They do optimization with the QE HF-X
They do some ultrafast targeted assays.

They find that if they've got 9 min per run they can kick out 84% accuracy in diagnostics.

And here is the fun part! If it's negative by PCR but positve by LCMS who are you going to believe?

Something that heats up fragile nucleotide strings and stuffs things between the strands, cools down and repeats?

OR

Something that measures a fundamental constant in the fucking universe (mass to charge ratio) directly. I dunno. I guess I'll chance that PCR thingamajing? Who doesn't love an indirect measurement?

Or something that says --
Negative (noise!)

And

Positive!! (Peak!)

Look, I know this isn't going to get out of my little bubble of mass spectrometrists that I preach to all the time, but COME ON.

People are doing spectrophotometry on colorimetric assay reagenst right this second in hospitals across the country to crudely measure one protein!

And here we are with cool research tools that can do so so so much more....

Maybe it's time for us to step up and throw slow, annoying, insensitive and inaccurate diagnostic assays to the side and just take over!!

Phosphoproteomics of SARS-CoV-2 infection suggests new treatment options!!

I'm guessing you've probably seen it, but if you haven't, you should stop and read this.

As a field, we've been messing around with phosphopeptide enrichments at least 12? 15 years? And -- yeah -- it's cool that we can pull down phosphorylated peptides and identify/quantify a bunch of them, and put them into big databases. Sometimes, though, it feels like we're just going through the research motions.

This is an actual real-life application of phosphoproteomics that shows -- it's actually useful. Not to put down anybody else or the thousands of publicly deposited studies, but this group just used our very standard research experiments to find new ways that we could potentially inhibit the replication of a virus that is fully confirmed to have killed 1,425 people in my country yesterday.

HOLY COW. Wait. Can proteomics be helpful to the world? Is this one of the tiny silver linings in this global catastrophe?

This blog is supposed to be more on the technical levels of how to do this stuff. Or just the ramblings of some guy who reads too many scientific papers while experimenting with the LD50 of espresso.

How'd they do it?
SP3 based digestion
TMT10-plex labeling of 125ug of peptides per channel (to 1.25mg total peptide)
250ug of the total peptide amount was fractionated (high pH reversed phase with a big (4.6mm Waters Xbridge column -- [I think the same one that Jesper Olssen's lab uses for the high resolution fractionations, but don't quote me -- I always use the 2.1 mm one, but I may need to scale up, both group's data might be better than mine...but both have better instruments....worth investigating....]) into 96 fractions over 70 minutes
The 96 fractions were concatenated to 24 for LCMS

The remaining 1mg of peptide was enriched with Fe-NTA and then fractionated differently. They used C-18 stage tips, fractionated to 8 and then concatenated to 4.

Two different LCMS methods were used on the Fusion Lumos --- SPS MS3 for the whole proteomics (Ion Trap on Turbo, MS3 at 50k resolution)

MS2 based quan at 50k resolution was used for the 4 phosphoproteomics files.

At first I was a little thrown, but the real reason that SPS MS3 is useful is coislation interference and ratio suppression. When you do a relatively simple mixture, I get a lot less improvement from MS3 over MS2.

Also -- it's phosphoproteomics. Do biological systems function with a 1.8-fold upregulatation of a kinase changing anything? I dunno, but it sounds unlikely. When I see a real phosphorylation event it's typically 5-fold minimum. A lot of the time it's a huge difference.

PD 2.4 for data processing. I'm a little lost on how the data was combined with ZScores, but Siri is telling me that I have work to do, so I'm gonna wrap this up.

So.....the phosphoproteomics points them to some things that they could inhibit to reduce the viral replication -- so....they

INHIBIT THEM

AND TEST IT

AND IT WORKS

Offset multiplexed targeting -- The brilliant targeted method you should have thought of first!

I have been dying to talk about this paper and it is finally out!

If you have been running a mass spectrometer of any kind -- get ready to feel reeeeeeaaaaaallly dumb that you didn't think of this first.

I'm allowed to say that I was lucky enough to be one of the reviewers if this, right? And this extra time has been very useful becaus my ego has almost recovered from never once thinking of this myself.

If your name isn't on that list of authors above, I'm going to now refer to you (and myself, of course, because I'm just as guilty as you are) affectionately as...I dunno...let's go with "Dumbass"

Hey Dumbass! You know how you can get a heavy labeled peptide for anything you want? Heck, you can even order complete heavy labeled peptide mixtures?

Oh -- you knew that? Good. So did I.

Hey Dumbass! Did you know that you can offset your fragmentation window?

Oh -- you knew that too? Yeah. So did I.

So....Dumbass....did you ever put these things together and think -- I could just put in my heavy labeled peptides and when those triggered, just offset my fragmentation backward to where the light peptides would be?

No? Neither did I.

I can't overemphasize how much I am going to use this technique. You can essentially bypass all the challenges of PRM with the addition of a heavy peptide standard that you target and then offset.

You need to do a little bit of math to work out your offset, if -- for example -- I just wanted to characterize a pathway of interest, all you need to do is get heavy peptide standards for that whole pathway. They make kits for these.

There are complete whole protein and phosphoprotein heavy peptide standards for entire pathways....

...as well as for critically important central proteins like TP53 which are almost impossible to monitor with traditional proteomics due to their exceptionally low expression levels! 

#ALSMinePTMs final data submission deadline this week!

We've had some great submissions for the #ALSMinePTMs Proteomics Data Mineathon Challenge and we're about to get them summed up and over to the judges.

If you haven't sent your submissions over because your life has been disrupted by some pandemic thing, we'll still work you in this week. If you're aaaaaalmost there, shoot an email over and we'll do what we can.

Submissions can be sent by DropBox, GoogleDrive, BOX or even summed up in a powerpoint, Excel sent to:

lcmsmethods@gmail.com

If I had to choose one webinar to attend this month? Stats in Proteomics!

Who spent 11 hours on phone calls and Zoom webinars yesterday?

Everybody!! 

Okay, if you're getting a little fatigued with all this amazing remote learning and need to be a little more selective about what you register for -- what's a big huge problem with proteomics that we can fix from home by designing our experiments a little more smarterishly? And analyzing them with some more robusterlificness?

Experimental design!

And next Tuesday there is a talk at noon courtesy of this cool organization that you can grab a salad (lunch where I'm at, and I assume that people in California eat salads for breakfast) and get a statistician's perspective on how to make your proteomics better.

This is free and you can register here.