Got a few minutes to read through a summary of 210+ references and a thoughtful perspective of what proteomics does -and doesn't - do well today?
now also at www.proteomics.rocks
Got a few minutes to read through a summary of 210+ references and a thoughtful perspective of what proteomics does -and doesn't - do well today?
I'm re-reading this paper from last spring and - wow -
I had another mandatory Microsoft update thing which 1) enabled their ....sub-performing.... blend of ChatGPT and Bing in the lower right corner of my screen where the button should be to get me back to my desktop... and
2) Makes it so that if I open a document from my desktop and go to save it as a new name it defaults to some imaginary "cloud" thing so I will never ever be able to find it again. You can "other save options" or you can:
Open every one of the MS Office things you use
Go to file
Go allllllllllllllllllllllllllllllllllllll the way down to Options
Go to Save and shut off this toggle that you never turned on in the first place.
You are welcome.
So...front end ion mobility things like TIMS are gas vs electric pull, right? So what happens if you run 5x the flow rate for your LC input? Will that alter the gas pressure enough to change your perception of the collisional cross section (CCS) and thereby change the observed 1/k0 values?
Really really cool study on this here!
You'll note they're using the ESI source, but if you're using the CaptiveSpray you're blasting your solvent directly into your glass capillary, right? I suspect that you'll see something similar when going from 100nL to 2 uL/min!
You know when you finally solve a mystery that is driving you completely bonkers and you get that relief at first that you solved it.
But then you realize how many samples you have to go back and rerun and you think "maybe I should really truly quit being a mass spectrometrist"? I'm having one of those days.
Quick solution.
If you upgrade your nice TIMSTOF instrument to TIMSControl 5.0 you'll get the thing above that I labeled #1 and - it is AWESOME. You get much better control over your mobilogram windows that can be guided by your real data. Using it right for DDA can give you dramatic gains in number of relevant fragmentation events. Run a file with your dumb mobilogram that you drew freehand. Then load that file into that PASEF precursor region thing and it will load your whole mobilogram average. Then you can draw the smartest possible mobilogram(s) for your actual experiment. It's beautiful and it might be giving me as much at 10-20% more PSMs on some of these samples. I can't wait to try this on TMT (which moves funny in IMS space and TMT and TMTPro are different, see figure 1C and 1D here)
However, you don't get this for free. You'll have two new glitches to deal with. The first one is sort of funny. Do me a favor and open TIMSControl 5.0 if you have it and turn on Stepping. Then turn it off. (That little toggle I marked with 2). Then note what happens to #3.
What it'll do is swap your fragmentation energies. Honestly, sort of harmless, but just funny. You'll find that you trigger the same number of peptides but you don't identify and of them because you hit them with the appropriate level of energy for fully liberate TMT11-plex reporter ions. BOOOM.
The other one is more nefarious and what I feel the most stupid about.
If you load older TIMSControl DDA methods what we find is a dramatic reduction in MS/MS events. If I run the exact same sample with a DDA method from TIMSControl 4.0 back to back with a brand new one with TIMSControl 5.0 while trying to keep everything the same,
TIMSControl 4.0 in 88 minutes ~25,000 MS/MS scans
TIMSControl 5.0 in 88 minutes ~140,000 MS/MS scans
What it looks to me like it is doing, but this isn't my job so I ain't gonna spend more time on it, is incorrectly reading the Target Intensity and Target Threshold values. Like your max target is now your min target.
I thought maybe it would be as simple as skipping a line in the method, but woooooooooooweeeee, the methods files are very very different.
What deserves two "very"s? There is a sum difference in 280 lines of the methods files between the two software builds. So...while I can't say for sure that the intensity and threshold are swapped, I can say FOR SURE, that the method files should probably not be used interchangeably at all.
This isn't meant to be a criticism of the very nice and incredibly small team that is building the software for all of these instruments.
Josip Blonder had one of the biggest effects on my development as a scientist in this field. He's got a fancy emeritus/pseudo-retired status at the NIH now, but apparently he's not just fishing off of Hvar, even if he isn't pulling down the surfaceome these days. He's updated Proteomics for Drug Discovery and it will be out in August!
And when I stumbled on this, I also found this one is coming out just before it!
The first book on single cell proteomics by mass spectrometry!
How cool does that sound? Zoom should be good for a huge number of external participants! You can check it out here if you're interested. https://jhjhm.zoom.us/j/99656763944
Whether you hate this plant or not, this is a really really cool new preprint!
Personally, my favorite part of this is TMT LABELING WITH AN OPENTRONS! What a nice cheap option for this step that I wouldn't have thought you could do (the OT-2 pipetted don't do a great job with solvent/positive vapor pressures). So while I don't actually have time to spend on this right now, I'm absolutely going to find out how they did this part, even if I don't care about this stupid plant. 😇I just prepped and ran an absolute shipload of plasma samples last week using a lot of the best of today's technology. S-Trap 96-well plates so it's fast and reproducible. EvoTips and EvoSep one so they're clean and - again - ultra reproducible and the newest best diaPASEF on a system with all the updates so it's screaming fast on nice little windows.
And......700-ish proteins identified. Which is just about the same f'ing numbers I'd get for plasma in 2011 prepping in a FASP filter and using a terrible old brown Eksigent nanoLC and running samples on an LTQ Orbitrap Velos system. Give or take a 100 protein groups. My memory has faded over the last 13 years.
The point is that it's generally pretty easy to see those top few hundred proteins in plasma and it doesn't seem to matter all that much what instrument you use. You're still stuck at the top few hundred.
But we've been seeing people getting 4,000 proteins from plasma recently. Not the completely unproven to have any quantitative accuracy whatsoever aptamer based stuff, but with super smart sample prep and runing on the newest and best instruments in the world.
The magic for LCMS based proteomics of plasma appears to have been solved. You just need ...$1,000 per sample and a TIMSTOF HT or an Orbitrap Astral...... and most of us have none of those things.
What if you don't need them? How would that change EVERYTHING?
You might recognize some of these names as being the most annoying nerds about "quantitative accuracy" and "matrix effects" and a bunch of other annoying things in proteomics. You'll find their most annoying traits in full force in this manuscript. Translation - holy fuck this looks like the real fucking deal.
The prep looks a little annoying, but it sure beats most of the ways we need to get past those 700 plasma proteins (deplete, run 74 offline fractions) and they conveniently automate this magnetic prep with a generally affordable KingFisher robot. All the scripts to automate it are already available and this Github is the fastest way to get you to the actual RAW data.
However, since this is a MAG-NET based method I don't see why you couldn't use other convenient things
I don't have a KingFisher but who hasn't experimented with automation on an OpenTrons and has one in use or being used for pipette tip storage somewhere. Drop in one of these and I don't see why you couldn't do this whole prep.
Are you interested in the fact that it's way way easier to see metabolites in single cells than to quantify proteins? Then you should check out this great -and short - new commentary!
One thing I learned in my short terms running Metabolomics services is that almost no one cares about all of the things that we can see in an untargeted run. When someone wants "Metabolomics" they almost always have one pathway that they actually want. They really want glycolysis or they really want this one set of molecules that are a side effect of the electron transport chain, etc., While running global is by far the easiest, you are always stochastically sampling those convenient metabolites for MS/MS and so the ones they're interested in are often not the ones where you have your highest confidence identifications.
What is really really super helpful to know up front is - what do most people actually want? That way you can build a panel or panel list that someone can walk up and pick off of a menu. I bet people who are good at Metabolomics with years of experience have those.
This group is very clear on what they consider would be the highest priorities for single cell metabolomics!
Some of these are easier than others - I like succinate (we use a heavy in our dilution buffers as one of our internal controls) and I'm cool with CoA metabolism. Nice big molecules that many retain without miserable chromatography). Others are harder, but you'll never find a nicer list of priorities for a rapidly emerging field. 100% recommended!