An Exploris pushed to 2 MILLION resolution!

I will lead with the fact that this appears to be another study where Dr. Makarov was in his lab having some fun with his somewhat famous invention, rather than a product we'll see soon. 

There is, however, some precedence here with something that looked fun later becoming a reality. The closest example obviously being when he pushed an LTQ Orbitrap to one million resolution using a hand picked Orbitrap and some painful manual tuning, and now you can up and purchase a Fusion set up in that way, so you never know!

In this study the team pops a hand picked Orbitrap into an Exploris 480 and does a lot of fine tuning through the internal development package. It doesn't look or sound like fun to me at all, but they get over 2 million freaking resolution and show what you can do with that much resolving power. This is a short and fun read that I highly recommend. 

Theoretical considerations of "next generation" proteomics!

 Now for something completely different....

Non mass-spec based proteomics is coming, or is here already. It's not going to be perfect, but LCMS proteomics isn't either.

Magnus discusses two such technologies and presents some surprising results. A high number of canonical human proteins can be identified with knowing the correct sequence of just 3 amino acids. Of course, the canonical human proteins are only part of the story but a lot of LCMS proteomics only considers those sequences, making it a fair comparison. 

Neat integration of TMT phosphoproteomics with transcriptomics!

The effects of my over-exposure to Arabidopsis from my years at an agriculture focused universities might finally be fading a little. Or this great study from last fall beautifully demonstrates something I've aspired to pull off myself and never have this well, and if you can pull it off in this awful plant, you could apply this to interesting things! 

If you don't want to read the preprint from the figure at the top, here is the hook.

There are a lot of studies out there that have brought together proteomics and transcriptomics. Most of them have required a bioinformagician on the team to pull it together using a series of tools that I am probably not smart enough to install.

TMT-NEAT, however, is a perfectly packages Shiny web application that you point at your MaxQuant output files and it pulls your data together.

To this day, virtually no tools out there autonormalize your phosphopeptide abundance to your protein abundance. That's why this lazy tool I made gets downloaded from the blog so often. TMT-Neat does this directly from the MaxQuant output files and does it the right way, not the lazy way! 

Honestly, I'm guessing we haven't seen this paper in print yet because it's going to be in a big journal soon and I don't even think that what I'm impressed with is why it will publish so well. The majority of the paper is focused on what they actually found, which seems like kind of a big deal. I hope the reviewers at least appreciate the informatics. 

TMTPro16 vs TMT18!

Well...this post wrote itself! 

This new paper compares TMTPro (TMT16-plex) vs the addition of 2 new channels! 

To compare they labeled some cells with TMT18 and TMT18 minus 2 channels (because that's the TMTPro 16!) and compare them. As you'd expect, virtually the same results, but with more channels! 

If the 17th and 18th channels are for sale, I don't see them up on any websites yet, but I bet they're coming! 

Shoutout to Dr. Robitaille for the heads up, and for the better joke. 

IsoPlexis -- Targeted single cell proteomics panels! No mass spec required.

When I met with some applications scientists from this interesting little company, I told them that every time they referred to their instruments by the term "Iso" it made me involuntarily shudder.

Last year I was fortunate enough to lead a very small team of very smart people through a full multi-stage ISO17025 and then government audit process. There are exactly two ways to get multiple government bodies to bless Q Exactive assays. 

#1: Put this flowchart on your wall and live by it: 

#2) Put this exact same flowchart on your wall and use dry erase markers rather than more ...permanent... markers. (Why do I have so many pastel markers?) And live by it. 

Name aside -- what's an ISOPlexis or ISOSpark?  That link will take you to their very pretty website. They use thicker border lines on shapes than even I do. 

It's possible proof that I should shut up and listen more, because it is a surprisingly powerful platform for looking at proteins in single cells without needing to master a 2,000 pound instrument. 

What's it do? 

It uses little flow microchips to sort a small number of cells (you typically start with 30,000) by basically their normal propensity to migrate and flow around into little microchannels. An imager looks through and discards empty wells or wells where more than 1 cell ends up. It's pretty efficient with something like 800-1200 wells ending up with one cell in them. Blank? There is your background subtraction. More than 1 cell? Ignore. 

When the cells are stuck they're lysed (if the workflow is for what is inside the cell) and fluorescent antibody things are used to quantify the protein signal. 

What's the catch? Well, it's antibodies, so you've got their variability to consider. And you're obviously limited by the number of antibodies, dyes, fluorescent thingies, that you can use at once. 

However, they have panels that can quantify up to 32 proteins per cell! If your cells distributed well, 32 proteins per cell across maybe 1,000 cells? That could be very useful validation data. You can also work with them to build your own targets if they don't have a good panel for you. 

The instrument is less expensive than a mass spec (unless you're talking Advion Expression) but the chips aren't free, but they're WAY less expensive than doing 1,000 western blots, particularly if I have to do them (...which probably means 75,000 attempts....) 

ISOPlexis also realizes that we're out there (the ones with the half-ton machines) and they seem interested in bringing us all together. Last week they sponsored a really cool proteomics webinar with GenomeWeb where some great people from our side talked about where we are with our single cell -omics plans. I didn't find out about it until it was going on, but it's on demand here! This link might be glitchy because it's linked to my name, but I think it will get you there. 

Hey! If you do decide to check this technology out and want to share data, please let me know. A demo to run 2 samples on this platform costs $10,000!!  I'll ask for money if you spend it and aren't really mad about it.  I'll leave out any conspiratorial things about an how an unproven technology might not want somewhat established proteomics people to see their data. They just had a successful $134 MILLION DOLLAR fund raise! 

SILAF -- Efficient and inexpensive whole organism labeling!

We occasionally still hear (here? come one, brain) about Stable Isotopic Labeling (with) Amino (acids in cell) Culture (SILAC). Which is probably still considered by many to be the gold standard in proteomics quantification.

There is still a ton you can do with SILAC, but when you first start planning an experiment one of your first impressions will probably be

It ain't cheap. Particularly when you consider how long you have to feed these cells really expensive medium. Or feed really expensive labeled organisms to other organisms for whole organism SILAC labeling. 

Solution? Ditch the C for an F, you! 

This new study on SILAF demonstrates an elegant labeling method for C. elegans. Crap. That didn't work. A...nope.... Nothing works with drosophila. whole organism using fruit flies.

Important note that I didn't know until I read this. You should totally avoid heavy arginine labeling, anyway! Check the references.

This group uses heavy lysine food and digests with just LysC and gets fantastic efficient and much less expensive full integration into the fly. It looks to me like this can be applied to any other organism model. 

I was really going to use some Jeff Goldblum joke, but I'm still bummed about the nematode. Maybe later! 

Reversing stem cell state differentiation!

Warning: This is a stem cell paper. In order to read it you must get a spare sheet of paper on which to write all the acronyms and tape it to the side of your monitor. It's like reading an LCMS proteomics paper if you haven't been running your own system for 3 or more years. 

I don't get all the biology, of course, but I kind of get the gist of it. Primary cells in culture get old and you can't use them anymore. This group has a way to push the cells that they're interested in back into doing what they were originally doing before they got old. Super cool, right?!? I can draw all sorts of ignorant implications from this and I'll keep them to myself.

Okay, while it's cool to be able to do this, the next obvious question is: what is the responsible element(s)? Can you hunt it down and put it in a vial? 

Using a combination of RNASeq and proteomics (Q Exactive HF) they come up with a list of potential candidates. The RAW and processed files are up on ProteomeXchange at this identifier: PXD021069

WHAT? Single LIVE cell NATIVE intact protein mass spec??

I'm super impressed. All the pieces to do work like this already exist but you have to pull techniques from all over today's best analytical techniques to put it together. This team did and...wow...

The single cell metabolomics people have live cell probes where they use a microscope and run a tiny object into the cell, pull out the cytoplasm and direct infuse it. It looks amazingly difficult to do right, but I know a lot of work is going into making it easier. The last couple talks I sat in on made it seem like something I probably couldn't do successfully rather than something there is no way in the world I could do successfully. I mean, you've got to get a cell and then stab it. 

How do you even find it? 

Super fancy microscopy or something. 

Not to put down the metabolomics people but small molecules are pretty easy to electrospray. You go to nanospray and you've got plenty of signal to work off of.

You know what isn't easy to ionize with electrospray? NATIVE INTACT PROTEINS. For native I like about 10 more protein than with reduced proteins. 

The fact that this group can resolve native hemoglobin from a single red blood cell is just nuts. Yes, there isn't much in an RBC but hemoglobin, but those cells are tiny. 

For the LCMS work you need to go to the supplemental where you really see how much optimization went into making this work. The instrument used was a Waters Synapt G2 HDMS. 

FAIMS Pro for Top Down Proteomics!

Okay, so it obviously isn't revolutionary or anything to think that FAIMS is good for cleaning up your top down background and boosting your signal. Where do you start, though?

Right here! 

Okay, so for those of us who might not be able to upgrade to the newest instruments, I would like to direct you toward an extremely interesting submission at PRIDE from some Thibault guy that I've heard knows something about FAIMS.

The submission is called "Integration of FAIMS Pro on a Q Exactive mass spectrometer" and you should search the files, they're impressive. 

An amazing new automated platform for single cell proteomics!

Hmmm....I wonder what Mechtler Lab is up to these days with 3 single cell papers in as many months, and a single cell proteomics conference coming up right around the corner in Vienna? 

This new preprint is ridiculous in a number of ways. First of all, what is even happening with the sample prep above? I pulled the links to the hardware they're using and it looks to me like this thing is doing the sorting, and they go right into a stream for digestion labeling AND injection to LC? 

The LCMS is an Exploris 480 with FAIMS and surprisingly they appear to use a trap column, which -- in my hands -- has been disastrous for my signal, but I pulled some files from ProteomeXchange and these files are intimidatingly good. 

E4 fragment ions?!? Geez....

Okay, so this is hands down the best SCoPE-MS data that I've seen on a D20 Orbitrap. 

There are no tricks here, they compare NORMAL SIZED HUMAN CELLS, (shoutout to anyone out there using 7 foot long giraffe neurons for single cell.) 

This sample prep method should be investigated! 

INTERNAL fragment assessment in top-down proteomics!

 I'll be honest. I don't fully understand this cool new study. What I do understand is that when I fragment an intact protein I get fragment ions from the N and C terminus(es) and only in the very best cases do those fragments extend to the middle. 

What they've worked on here is alleviating this by optimizing parameters and considering that the fragment ions might not be working all the way from each terminus. They use a special Mascot version (TD?) but explain how you can process this data in a similar way using other tools with a link to an older study (direct link to that study here). 

It makes a lot of sense to me, we're hitting these intact proteins with a lot of energy and why wouldn't we get internal fragments? Thinking we're exclusively seeing b and y ions is convenient, but we see secondary fragmentation in some large peptides and definitely in peptides with PTMs, so why not proteins? All the work is performed with a Bruker FTICR so they use some pretty amazing mass tolerances to demonstrate that they can actually observe fragment matches. 

Assess "missingness" in DIA data with OptiMissP!

Missing values?!? Isn't avoiding those the whole reason we even do DIA? I thought there weren't any?!?

Yeah, there still are, particularly in really complex matrices. If you want to assess the levels of it, you can manually search or you can use -- OptiMissP! 

Don't feel like reading? Me either! (Just kidding, it's a pretty cool paper.) 

You can get all the code at github here. 

Or you can just mess around with their great Shiny app here! 

A nice preprint that breaks down the Exploris'es

This was not what I was looking for at all, but it is interesting! 

What's an Exploris 420, 680 and/or 102? What makes them different? 

This preprint from June has all the schematics and some data showing what the different models can do. 

Donkey milk proteomics!

Did you think about donkey milk today?  Me neither! And I wouldn't have guessed it was useful. Turns out that it is: 

1) Good for children with allergies to cow milk

2) Has a long history of being utilized for various medicinal purposes.

The first seems less fantastical than the other, right? 

This group went out of their way to study the donkey milk proteome and reveals an interesting story. The start off by doing intact proteomics, and double back to fill in the blanks with LysC and Tryptic digestions. 

1) They explore both the proteins that could explain why people who are allergic to cow's milk may not have the allergy to donkey's milk

2) The antimicrobial properties? They totally confirm it. You can even leave donkey milk out for days longer than cow's milk. Antimicrobial super powers? Sure looks like it and this group points at what might be causing it. 

Altogether it's a really nice study with some cool biology explained by some really good mass spectrometry. 

Were you dying to know about the VAMPIRE BAT serum proteome?!?

You know, prior to this recent little thing that happened with a bat virus  I might have cast this paper in a different and less serious light. However, it seems like we should probably know everything we can possible know about anything. 

Therefore, I present a no-nonsense blog post about this great new study. 

The study focuses on vampire bats in Belize and samples come in as part of a longitudinal study. Rabies is involved and so is livestock due to agricultural alterations of bat habitats. Yikes. See. Totally serious stuff. 

All LCMS is performed on an Orbitrap Fusion 2 "Lumos" using DIA. As you might imagine the vampire bat FASTAs may not be as well developed and annotated as some other organisms, so the informatics section...

...better just skip that part. 

Do you want to know if they found high quality MS/MS matches to peptides that match the middle eastern coronavirus? Or should I ...lay this one to rest... already? 

Predict peptides with PROSIT and MSPiP for proteogenomics!

Wow. Okay, of course this is one way to do it! Actually, it seems blindingly obvious to me right this second, but I certainly hadn't thought to do it the way they did in this new study!  

There are several smart ways to get around proteogenomic challenges that we're increasingly being encouraged to think about. It's a lot easier if we can just use the UniProt SwissProt little tiny library of protein sequences, but they ignore:

mutations,

individual biological variation,

the fact that some of those sequences might have been Craig Venter's and if you were picking a normal human being to pull sequences from, would he really be the best pick? 

What this group does? They pull out their little nanopore gizmos and they do some sequencing. Instead of feeding their Variant Call File (vcf) or whatever is the same thing in Nanopore world into a search engine, they feed those sequences to deep learning peptide fragment and creation tools. 

BOOM! Modeled MS/MS spectra that they can match against theoretical! Smart, right? 

Did they use MaxQuant.Live targeting in reverse to boost Ub and SUMO coverage?!?

 I'm mostly sure I understand what is going on here? Maybe? 

What I think is going on is that I just didn't realize that the super powerful MaxQuant.Live targeting function could be used for ultrasmart exclusion as well. And if that is what is going on 1) I feel kind of dumb for not realizing that was possible -- I could totally use this! and 2) you can use it to dig deeper into your data and find the really annoying PTMs like SUMOylation! 

Okay...but I have the interface (Beta 2.0! With Exploris support, which clearly couldn't be used to get into any other instruments with the same handy universal interface --

)

open 

And I'm not sure that I could reproduce this on our instruments without...sigh...reading more... 

This post appears to have a theme now. I'm not sure why.

If you are also stymied by a number of options that seem too challenging for you to get away with just tryng them all to see if they work, I bet the secrets are in one of these places.

The MaxQuant.Live paper? 

The MaxQuant.Live documentation pages? Oh. Maybe I found it here. Not 100%, but it narrowed it down. 

The google group page?  It's here somewhere! 

Unrelated.

It's finally time for the Journal of Proteomics Research special edition for tools!

 I've been looking forward to this JPR special edition since I finished the last one! 

You can check it out here! 

Just flipping through:

Updates for Percolator, a new glycoproteomics engine, python tools for TIMS data (holy cow, does the TIMSTOF need more functional software) an open library for metabolomics, and a bunch more R packages. 

Get the right tool for the job! 

IonQuant paper is out!

 

There is a decent chance given the amazing amount of traffic on the MSFragger Google Groups channel that you've already tried IonQuant and may be actively interacting with the developers to make it better and faster and more powerful.

If not, you should check out this ultrafast label free quantification application which actually works out a metric for false discovery rates for match between runs. We're literally using it every day! It's how we process our daily quality assurance runs on TIMMYDuncan.