Wanna solve the bee problems? The argument for true multi-omics!

 

Even Doctor Who was like "that IS weird, innit?" when asked about the global declining bee populations (it's the pesticides, some of which are now just as abundant in the air as our friends the polysiloxanes)

But declining populations aren't all that we need to know about our stinging little friends, and this very thorough review suggests doing true multi-omics to figure it all out.

Really cool, but somewhat strange read, because you'll go a few paragraphs and then remember that it is supposed to be about bees, but then they'll mention them again on the next page somewhere.

But, for real, take it from someone who has done tens of thousands of sensitive and almost exclusively high resolution pesticide assays - we've miscalculated the half-lives and impact of very common insecticides. They're absolutely everywhere. 

WAIT. Coincidental with this new review is this new preprint on bee proteomics! 

Metabolic labeling in single cell proteomics dissects protein regulation!

 

Fuck. This new preprint is awesome. For real....this is really really good, and I saw some stuff about it on LinkedIn and assumed it was blown out of proportion. Its not. It's really really cool. 

What if you found a really good in vivo model that you could control well to try and dissect what and how single cells actually regulate protein abundance. There are amazing resources for protein turnover and mechanisms in bulk digests, but nothing comprehensive in single cells. I mean...we couldn't actually do that...yet...right? 

In this case they specifically narrowed in on mouse tracheal cells. Then they metabolically pulsed the mouse diet. Then they did single cell transcriptomics and single cell proteomics (with multiplexing, to get 4,200 cells) and worked backward to figure out 

What proteins correlate with transcript regulation, which work backward to protein regulation and which are regulated by clearance (what I'd call turnover, but they make a point that in vivo there are more than one way to drop the abundace of the protein in a cell. 

It's the kind of basic science stuff that you know needs to be done with single cell proteomics so we can take that step forward and do applications - but...funding mechanisms are all pretty heavily biased toward actually applying it. 

For the mass spec side, multiplexed DIA was used...there might be some details missing in the method section for the labeling but it's based on a previous referenced work. It's essentially mTRAQ with DIA. 4,200 cells with an average depth of 800 or 900 proteins. 

I gotta run, but I'll be revisiting this one later. 

How mass analyzer resolution changes across mass ranges - and how to monitor it!

Due to the dispute between Guugle and Virginia Tech that resulted in - finally, I guess - the loss of my email, I also lost a lot of files linked on this blog that were still in that Googl drive. 

I'm hoping to find them at some point and put them back up. If you do see a link that is down you could try leaving a comments (it's hard to get through spam) or email me at LCMS methods@gmail.com (remove the space).

In basically all mass analyzers, we set the mass resolution at a particular point in the m/z range. The easiest example is always this one (you can click to expand) 

Note that at 558.3010 the resolution is nearly 50,000 arbitrary units (not arbitrary) and at 1273.0643 we've decreased to 31,700 resolution. This is an Orbitrap analyzer and this is how it goes. I've got boring slides on why (which...probably....have a dead link here as well....on the list now)

A funny thing happened around 2013 when every new Orbitrap magically had a lot higher resolution when marketing people thought "we should start reporting mass resolution at 200 m/z instead of 400 m/z!" Because you get more resolution as you keep going. On a GC-Orbi that can scan down to like 30 m/z you can get close to 1 million resolution! So the scan above was probably 60,000 or 70,000 resolution in the method (at m/z of 200) and this is image above is what you get at higher m/z.

What's less predictable are today's Time of Flight (TOF) analyzers! And this is funny because if they were purely just TOFs they'd be exactly the same resolution all the time. Ions go from point A to point B and that resolution is derived solely from the distance between point A and point B. This is, at static temperature, the same length.

However, in all of today's instruments they're not just TOFs. They're all basically hybrids in one way or another. They've got quads at the front of them and/or quads and traps and quads and traps and ion mobility devices and more quads and traps. Throw in a mirror or 7 (so your TOF tube isn't 25 feet tall....) and you've got a lot to consider here. 

The net result is that you've probably got to get those ions and accelerate them into the final TOF mass analyzer. And sometimes you've got to accumulate them and accelerate them. If all the ions were starting at the same spot at the same time - yes - you get the same resolution across the whole mass range. But if you've got a time of flight effect before you start the actual TOF analysis (through your quads and traps and IMS, etc.,) then you may need to make some compromises and every hardware configuration and vendor is a little different.

Let's take the one I know best right now, which is the TIMSTOFs. My first TIMSTOF had a reported resolution of 60,000. Which would be great across the whole mass range. 

But you don't get that. This image is stuffed into the supplemental of this paper somewhere. 

This is the very best separation I could get on a TIMSTOF Flex in 2020. This involved manually tuning the mirrors and tuning in the now-retired(?) OTOF control software. The trick is to manually override the mass that you're tuning on, but if you go too far one direction you can totally f' up the other side. I think this was tuning on 622 instead of the default (at the time) of 1222. So....you can ...just get that TOF resolution that you need to do TMT quan. 

DISCLAIMER - If your TIMSTOF is in a basement with water and HVAC issues, you won't maintain this resolution for very long at all. Not even 2 days. And that's the very very best separation I ever got. Realistically I could not resolve these peaks at 50% base height on normal experiments. 

The TIMSTOFs have gotten higher resolution. They'll get 80,000 at 1222 during install. I've taken some looks and - nope - I still can't reliably separate the 0.006 Da of the TMT reporter ions.

Okay - so that was a lot of words for - Orbitraps decrease as mass increases, TOFs have an ideal spot and you can tune them in one way or another to optimize that spot, but there are consequences. The best I can tell, they don't seem to change in a quadrupole - and I forget what it's like in an ion trap.

Time for a blast from the past blog post! Check this out! What if you could easily - intra-experiment - regulate your mass resolution on any instrument to see if it is what you need when you need it? This paper details simple reagents that could be used as pre-run or even intra-run controls! 

Flexynesis - Integrate all the -omics with deep learning?

 

I, for one am glad to see that when I'm not the only one who is at a loss to illustrate deep learning models and resorted to a commercial AI to make some squiggly lines (for the github tutorials).

You probably want to read this first before starting to install it!

Is this the thing that can take my metabolomics and proteomics of the same exact 800 or so tissues and integrate them properly? ...Not out of the box....

The current input categories are data that is already processed through R packages such as PharmacoGX and the single cell transcript measurements were processed with Seurat. This latter I find compelling enough to press the publish button, however. Thanks to work from Laurant Gatto's group (which should be out by now, I think, preprint link in here). I can get my single cell proteomics data into historical single cell formats. In my notes from a workshop he taught at EuBiC I wrote that he said "there is no reason we couldn't use this for other large -omics data" and also "omg, I'm way too dumb to be in this workshop" which I don't remember him saying, so was probably some level of intraspection on my part. (I miss when blogger had a spell checher embedded.... 😉) 

Adjusting single cell metabolites to match actual conditions?

 

I'm largely leaving this one for me, but for proteomics people take a think about how value neural networks (like DIA-NN) have helped us in distinguishing signal from noise in low level peptide data.

They really are evolving. We constantly do true/false by lying to the neural networks about where and when we have blanks and controls and non-human samples. And even 3 years ago, they weren't nearly as trustworthy as they are now.

Now...consider the metabolomics field which doesn't even have a good way for routine false discovery rate estimation in their data....then drop your signal down to just above your noise....

This thought-provoking new study tackles this and tries to really optimize their libraries for single cell loads. 

I'm clearly not qualified to really review this study, but I did enjoy it and it made me worry about some upcoming work we have intended to target peptides from single cells where we won't benefit from finely tuned machine learning/neural-networks to tell true from false....Makes me think we'll need to be even more skeptical of each peptide retention time, ion mobility and fragment distribution pattern than usual.... 

More crosslinking news! ChromaQuant!

 

Looks like a nice solid workflow (and Shiny App!) 

Got a fleet of (Thermo?) mass specs and 1 operator? HELIOS to the rescue!

 

Hey you! Did you trick a bunch of ...less informed....investors into buying you a bunch of mass spectrometers because you did genomics a while back? 

Did you already try underpaying a lot of very young and capable seeming scientists with chemistry degrees but no experience, and are SHOCKED to discover content level expertise is required to do mass spectrometry based -omics? 

After all your ...celebrations...that you were now a field leader in proteomics or metabolomics or mass spectrometry of some kind.. do you have just enough money left to finally hire a full time expert for that fleet of instruments? 

You'll be sad to learn that a fleet of instruments may take more than 1 person to operate. I personally think a 1:1 ratio is generally pretty ideal. However, if part of that money came from guys with tattoos on their necks and a lot of muscles despite the fact they don't seem to go to the gym much, you might need to get those big heavy boxes generating ".RAW" files.

HELIOS to the rescue! 

This might not be what these authors intended for HELIOS, but it might be where it sees the most use given the 10 or so "multi-omics" companies I have on my watch list for their upcoming bankruptcy sales. I could put a TIMSTOF HT to very good use if I could get one for a court mandated 20% of original price even if you're new at this stuff that you paid list price for it. 

Disclaimer - This blog post is a joke to promote a really incredibly clever paper from one of the very best teams of mass spectrometrists on planet earth. Nothing else written here should be implied as linked to any real events or your scammy company or any other scammy company. Heck, if you have a scammy company like the one I joked about here, I certainly don't know about it or 7 others. Don't tell me about it, shame on you, and go hire some professionals and do good proteomics for the good of mankind! 

Is this finally - in cell crosslinking -for real?!?

 

Okay - this has been going on for a while now - you might find 10 posts on this blog about the quest 

of protein-protein intereactions. 

Putting protein crosslinkers in to bind proteins together for proteomic analysis! 

Is this it?!? It looks pretty smart. But I also have a bunch of crap in my freezer that works in a centrifuge tube with 4 proteins (sometimes....) that absolutely does not work in-cell. 

If so, I'm 

(...ugh...you work with what you have when you've got 5 minutes for a blog post and a protein bar...) 

Boost that signal by lowering that FAIMs temperature (and resolution)!

 

I've been hanging out with people for the last year or so who are building super high resolution ion mobility systems (at least resolution of 300 while operating, which is significantly higher than any of my TIMSTOFs). So it's funny to see benefits of taking a low resolution IMS (10?) and making it even lower! 

Anything not getting trhough the FAIMS is going to waste, so this makes a lot of sense. Ultimately you're probably just using FAIMS to get rid of the +1 junk you can't sequence anyway and the PTMs that are much easier in downstream analysis to just ignore. If you can do that while losing less of your proteotypic peptides - BOOM - more signal! These authors demonstrate it adds ID#s in both standards and actual single cells! 

Deep visual proteomics tackles colon organoid transplantation!

 

Transplantation of organoids serves as a model for understanding the complex development cycle of the the colon, including cancer, and may eventually be a mechanism for building back healthy tissue.

In this stunningly beautiful study, 

this group dissects the process - often at a full single cell level - with striking depth and with cell type-classification. 

Single cell phamacokinetics figure I really like from last week

I've been presenting on this stuff for a while - and I don't think it's a one off observation. A professor in the audience (thanks Vinnie) provided an idea that might help me finally crack this one. Now...gotta talk someone into doing a bunch of cell culture..... Regardless, it's my first use of this particular meme and I'm relatively sure I did it right! 

https://pubmed.ncbi.nlm.nih.gov/38014353/

ECD of small proteins on chromatographic time scales?

 

If you're familiar with ETD (electron transfer dissociation)

And the much newer and (magical - I still don't get how using large magnets forces more democratic fragmentation, but it's fast as heck, EAD (electron something that starts with an A dissociation)).

Here is something called ECD (electron Capture(?) dissociation).

Which was equipped here into an Agilent Triple quad wearing an extremely tall tophat.

Edit: I have been informed that is a Q-TOF. 

This appears to allow top down proteomics of little proteins with ECD that is much faster than ETD! It still seems slower than EAD, but might be an attractive solution for upgrading an Agilent top hat in some labs? 

Finally the paper to convince you to toss that nanoLC off a roof?

 

If you've ever been to this blog  before, you might have encountered the fact that I despite nanoflow liquid chromatography. 

These authors seem to agree! 

I rarely quote the text of a paper completely, but I'm going to do it with this great new one. 

Begin quote: 

When I started my first proteomics experiment in 2007 or something, I thought "man...these motherfuckers are dumb....this is the shittiest 'separation' I've ever heard of". But you had to do it...probably...because the mass spectrometers needed 4 kilograms of protein separated over 6 weeks of chromatography time to detect 3 proteins besides albumin. 

Now that mass spectrometers don't completely suck, maybe, just maybe, we should go back and stop doing shitty unsustainable chromatography that only has the benefit of extremely poor reproducibility? 

End quote. 

Mass spectrometry proteomics - From single cells through new(!) clinical applications!

 

I'd seen this one months ago, and didn't dig into it until I had reason to send out some updated reviews. If you're trying to catch up on proteomics as of February 2025, it's worth the read.

15N targeted assay development and deployment on a fast ion trap!

 

I totally hoped this was going to be an untargeted plasma proteomics at low resolution/high speed paper, but that's okay, it's still pretty cool.

Now...if the price of this fast little ion trap would come down to an appropriate price range... we'd be in business. Do you want 5ppm mass accuracy on a benchtop Orbi that's WAY slower? Or 15ppm mass accuracy on a TOF that is the same speed as this cool little ion trap? Or is 800ppm an okay trade off for fast and sensitive?  I'm not willing to pay the same price (or more!) for unit resolution, and I know I'm not the only one. If they eventually price this thing in line with the instrument capabilities in a competitive market, I think they'll be all over the place! 

Proteomics in Forensics week!

 

I'm going to leave this here. I think it's a really intesting study, but I'm a little surprised by the conclusions. Caution is always good, and this paper is all about limitations of blood (spot?) forensics. While machine learning classifiers appear underpowered to make accurate determinations, I really wonder if a targeted analysis of specific proteins and peptides - in this same set of data - would come up with the same limitations.... 

Also - in case you haven't seen it, last week we dropped a new podcast that was also on Proteomic Forensics (in wildlife!)! Get it wherever you get podcasts and things.