Halloween TMT CageMatch -- MSFragger vs Proteome Discoverer +Chymerys!

 

FragPipe can do TMT stuff (as long as you aren't using a TIMSTOF or ZenoTOF) and it has totally looked impressive so far. How would it do against Proteome Discoverer with the fancy new Chymerys AI thingamawhatsit? 

Here ya go!

In the end, it looks like both do pretty well in terms of overlap, maybe the quan is a little better with FragPipe and maybe the number of IDs are a little higher with Chymerys. 

What the authors are concerned with is time for big cohort studies and this is where FragPipe really shines. FragPipe is massively faster here. 

Just to be argumentative, I'm still just here running MSFragger inside of Proteome Discoverer and doing my TMT with the IMP nodes so that I can get MSFragger speed (almost) and get all the visualization tools I want (and all of that is free stuff unless you're in a company). I do plan to finally check out Chymerys cloud soon and I got a free demo to check out when I find time. 

Lipids, glycans and peptides from one FFPE tissue -- by MALDI?!?

 

Gotta move fast because the life of a nontenure track PI is a unique pairing of exploitation and exhaustion diluted in carbonated personal financial insecurity. However, I think I have enough caffeine in my system right now that my brain is mixing that up with the joy I once held for life and  I'm going to use that to get some amazing papers off my desktop. 

How bout this one? 

SEQUENTIAL analysis of 4 separate sets of markers from one FFPE tissue with spatial data intact cause you used an imaging MALDI approach! 

Linear ion trap data independent acquisition for low level samples!

 

Crap. I thought I had this niche locked down where I apply alternative mass analyzers with superior characteristics to the few things Orbitraps aren't the best at. 

This group fired up a linear ion trap and did some low load DIA analysis. Ion traps are always going ot be more sensitive than Orbitraps. The ions strike the detector in the former, and pass by the detector in the latter. 

But DIA? On an ion trap? Sure! Why not? 

My first concern would be specificity, but that's always my concern with DIA anyway. You lose some mass accuracy, but if you're going faster can you make up for that with smaller windows? 

The nitpicky crowd of grumpy people who review for JPR would probably have really looked into inflation of FDR, etc., and not let a lot of bad data by. You aren't supposed to think those things, but there are some journals where you find out that you forgot to unlock one of your repositories during the peer review process, and some you only find out years later when some weirdo actually wants the files.  

The FAIMS Pro source they employed also totally has to help. 

I'm really would like to spend more time on this, but I'm having the sort of week where I have fondly remembered washing dishes in the back of an un-airconditioned restaurant in the summers for $5.25 an hour. 

Super cool work, and what a great idea for that linear ion trap you have sitting in your garage! 

How can proteomics help the processed meat industry?

 

Sometimes I just randomly click on things in my Scholar feed while I impatiently wait for caffeine to make my brain do stuff. Was it the espresso finally doing what I pay it to do or was it an abundance of new (to me) information in this new review that made me stop and pay attention? We'll never know. 

We all know about food adulteration, right? Like the fact that the world's total production of pomegranates can not possibly account for even a decent fraction of the "pomegranate juice" in grocery stores. What I've heard it that it is customary to introduce a pomegranate to red #13 colored corn syrup in the factory and allow said pomegranate to continue its holiday tradition of visiting industrial locations. 

In this study you'll learn of meat adulteration where protoemics has found all sorts of weird things in sausages. 

Also! Did you know that proteomics can help fine tune fermentation processes and that the during the controlled fermentation of meat the wrong microbe getting involved can lead to increased known carcinogens?  They describe some of these findings here. 

At some level, I think I assumed this stuff was happening out there. The US Food and Drug Administration buys mass specs for something, but this provides some insight into what regulatory bodies and food chemists actually do with the things! 

Great review on the application of higher flow rates in proteomics!

 

This new review is worth referencing if only just for the picture above. I think each vendor has an opinion on what is which different flow type and it feels like the cell phone speed thing. A "5G" one on vendor is just an arbitrary measurement to say it is faster than their last generation, but has no reflection on the speed provided by another vendor.  

I'm just going to just use this figure above from now on. 

Here is the review. It's in something called tAndf. 

Wait. While making the screenshot I clicked on the Listen button and it is reading the review to me! This is super sick. 

Jack  : Kyle, if I was in a wheelchair, would you visit me? Feed me? Brush my teeth?

Kyle  : Yes.

Jack  : Would you read to me?

Kyle : Why couldn't you read?

Jack : Just don't want to.

(Was gonna put in the link but watched like 3 minutes and remembered that old TV show is not appropriate for younger readers and this is, ultimately, meant to be an ultra family friendly blog.)

As these authors note, the success stories with micro-flow and analytical flow are limited as this point. Clearly, though, nanoflow is one of the remaining technical hurdles for mass spectrometry based proteomics implementation so getting past it could be transformative for the growth of our field. 

Blood brain barrier proteomics investigates the mechanisms of propofol!

 

It seems like basically every time someone starts really talking about how a drug works at a cellular or molecular level, you end up getting to a lot of question marks. When you start talking about drugs in the brain it sometimes feels like...

One of the important concepts that gets tossed around all the time is whether something crosses the blood brain barrier (BBB). What is that, besides the thing the words indicate? Please see gif above. Could you study how it works with proteomics? Again, please reference the 13th regeneration of the Doctor above. 

According to this nice new review it is a complex system of different cells working together, and you can there are multiple ways to study it which all have their own relative advantages and disadvantages (stole the summary table, click to expand.). 

One way to study it is to build your own blood brain barrier model in your own incubator, then you do crazy stuff like dose cells with different drugs, which is what this group did with propofol! 

I only thing that I associate with the drug propofol is the untimely demise of some pop music legends. And this synthetic blood brain barrier (BBB) model suggests that this compound is really bad for a number of different reasons. Adding propofol makes their BBB model more permeable and when they pull the cells for proteomics, these cells have some weird effects.

(Data was acquired with an Exploris 480 and data was processed with Proteome Discoverer. Results are on PRIDE as PXD033856) 

One thing they observe is a loss in H2AX after treatment. And that is one of the central proteins involved in observing and reparing DNA damage. After their proteomics they back it up with a western blot, which made me throw up in my mouth a little, but they probably had to do it to get a senior scientist to sign off on the paper and I've been told that I eat really weird stuff. There is other cool stuff in here, but I can't spend all morning reading about BBBs and throwing up because someone "validated" some HRAM data with rabbit blood and horseradishes. 

Wanna know what someone was eating 500 years ago? Do proteomics on their pottery?!?

 

I gotta type fast, but this is sooo cool!

Come on, did you ever think about how you'd learn about the eating habits of people in history with your mass spectrometer? 

They didn't waste a bunch of ancient relics, they worked on embedding albumin into pottery and got the solvent extraction conditions all the way they wanted them.

VIOLIN! 

Insight into the past! LCMS was done with an Orbi Elite (which are still totally badass instruments if you know how to run one, I don't care what anyone says [only being defensive because someone who will not be named made fun of my Q Exactive]).  

The Proteomics Standards Initiative -- Whispering the truth down an empty well for 20 years!!

 

A bright trainee new to proteomics recently asked something along the lines of "why are there all these different formats for all of these same things?" 

My default response is "proteomics people are a just bunch of assholes" and if I have time I go into the story of the most thankless committee to attach your name to in all of the history of science. 

The always ignored. 

Never cited. 

"Wait, we have a standards initiative?" 

Proteomics Standards Initiative! (PSI)

Wanna know what they've been up to? No you don't. You wouldn't be working on another spectral library format right now if you did. I won't even waste time putting the link here because no one is going to click on it anyway. 

Actually, this post is pretty short. Here is the link. 

I'm not going to read it either, I'm too busy working on a new universal mass spec format that allows the data from any instrument to be stored in a combination of wingdings and Emojis in .json format to facilitate the sharing of mass spec data over blockchain (literally the dumbest thing I could come up with, sorry if that is what the NSF funded you to work on.  I can't wait to review it!) 

I'll give you my summary the 20 year history of this initiative! 

Every year, this group gets together at a major mass spec conference to discuss the fact that we have 35 spectral library formats and 11 different "universal" ways of storing mass spec data and maybe, just maybe, we should have ONE. No one knows what conference this occurs at because conference organizers never remember to book a room for them to meet. 

The PSI typically ends up working in an unused supply closet or by a dumpster near the convention center and they still come up with great ideas to combat the biggest challenges that make outsiders think that proteomics people are completely insane or really really into drugs. 

They take these ideas and they schedule talks to present them that organizers forget to advertise, or are scheduled directly opposite Neil Kelleher himself. So the audience typically consists of one confused student in the audience who got the room number wrong and feels too self conscious to leave. 

And that's how you become a member of this unintentionally secret initiative! 

While this is possibly the third funniest thing I've written since I got up at 2am, I hope there is some substance in it somewhere. 

We do have a Proteomics Standards Initiative. 

We've had one for 20 years. 

They've somehow, despite every possible odd, made some progress in a field of obstinant vitamin D deprived individualists. 

Can you imagine what they might do if some of us paid attention? Maybe check it out? I'm not going to, obviously, but it would be cool if you did. 

Find those awful miserable stupid labile PTMs with MSFragger Labile!

 

While it would be amazing if every protein chemical post-translational modification would just stay put when we're gently converting them from solution to ionized gases with

and (for my fellow higher flow HPLC people) plenty of 

Some, maybe the majority of these wimpy PTMs fall right off! 

For a decade I've heard that MaxQuant can make use of diagnostic fragment ions and neutral losses and use it within it's data processing.

And maybe it can, and I'm just too dumb to get it to work. In either case, I don't care, because I CAN MAKE MSFRAGGER DO IT! And you can read about this here. 

If you're a nerd who can't keep looking to see what is in pull-down menus, you might have noticed this showing up in FragPipe 16 or 17 as the ADPr workflow. 

If you're working with ADPr I am truly sorry, because that modification truly fucking sucks. Is it gonna release an ADP? or an AMP for some reason (it might!), it sure is gonna neutral loss 3-6 times in HCD. And those stupid fragments ionize like pure arginine and your search engine is going to assume that you definitely aren't fragmenting peptides. Put some pure ADPribosylated peptides on your LC instrument, put the ADPr on the correct amino acid and tell me how many times a search engine finds it without you manually looking for it. If it happens once it is more times than I've ever seen it happen. 

UNTIL NOW. This weird Java program actually looks like it can do it!  (Low n and something I can't afford to put time into for a while, but this might be a huge step in the right direction.) 

Now, I can't entirely figure out how to modify this to look at another awful PTM, without a calculator and maybe thinking about it really hard, but this might be awesome!

Phosphoproteomics of C. difficile finds clear S/T sites!

 

So....this used to be Clostridium, but it is called something else now that starts with a C. that shouldn't confuse anyone, since it is still C.diff.? 

Regardless of the name it is still (one of?) the leading cause of hospital acquired infections. Go in for something else, possibly get C. diff. 

Here is a great study on this awful giant bacteria's phosphoproteome! 

These authors weren't even content with just doing phosphoproteomics (Q Exactive HF after Titanium Dioxide bead based enrichment) -- they also used/constructed mutant strains and analyzed those.

Lots to unpack here, including a better understanding of how an important pathogen goes into sporulation (which makes it resistant to most things you'd use to clean a hospital). And -- the unusual use of S/T phosphorylation sites. In school back when I went it was very clear bacteria used different sites, but that's what we're here for. Fixing textbooks and finding out ways to kill bacteria hanging out in hospitals. 

Platelet Rich Plasma (PRP) therapy proteomics explains contradictory results!

 

One of the most promising and controversial treatments for a lot of different injuries is the selected application of platelet rich plasma. I know a guy who is really dumb who couldn't get PRP therapy in his country so he does medical tourism to get treatments. 

The science has been confusing on this topic, including some completely contradictory findings. This sounds like a job for --

--a field that can't find any workers because we don't treat it like a field of scientific study! (That gif is from a crappy old cartoon called UnderDog). 

I mean, Proteomics!  You can check out this cool study, here. 

There are actually two main ways of obtaining "PRP" and they get patient samples from both and treat human cells of some type (human dermal fibroblasts, which I assume is a cell type that makes sense). 

What they found? HOW the plasma is enriched and how it is treated completely alters the cellular response and goes a long way toward explaining the observed contradictions.

Getting a better handle on this is critical in getting more insurance companies to cover some fraction of these expenses in countries where medical care is treated as an exploitational business model rather than a basic human right! Progress! 

New FDA fast tracked combinatorial therapy discovered by proteomics!!

As a field, we could use some victories to make the next step.

Next time someone says "oh yeah? what did proteomics do to really help someone?" I got you. 

CHECK OUT TODAY'S FDA ANNOUNCEMENT!!! 

Confused? Okay -- so until right before the pandemic, if you heard you had cancer with a KRAS mutation, it was bad. Real bad. Pancreatic cancer is primarily KRAS mutation driven. But Amgen did the unthinkable and targeted the untargetable KRAS mutant proteoforms. And based on their design (AMG510 -- which became sotorasib) lots of drugs are coming. 

Here is the problem. Like any drug on any complex thing resistance popped up.

But this group was right there working on it! 

They did exactly what you should do to study drug resistance. 

They low-dosed cells in culture and pulled proteomic samples over a long time course (for real, it's really beautiful work, this study has been on my desktop for almost 2 years and read multiple times).

The used a Fusion 2 for the TMT, but I (and now I forget) I think it is MS2 based quan (but maybe MS3 based for the TMTPhospho). But they came up with 3 really interesting proteins. One of which was SHP2. 

When they inhibited these 3 proteins -- cells lost their resistance to sotorasib!

That FDA link above? A combination of the SHP2 inhibitor (BBP-398) + Sotorasib. This study is only two years old and it led to a fast tracked combinatorial therapy to help patients!

If someone else found these resistance mechanisms first I sure haven't heard about and I feel pretty plugged in to this scene (who knows with pharma people and their secrets). But I think this is the study that found this! 

Proteomics (done well) can help people right now! 

FragPipe on High Performance Cluster -- here's how you do it!

 

I was somewhere a couple years ago and I had what felt like unlimited access to unlimited cores of data processing due to a lot of infrastructure on supercomputing (that wasn't being used a lot?)

My university meters supercomputer access and it's sort of expensive, so I don't want to spend time tinkering with the darned thing and burn through a bunch of cash I could be spending on trypsin.

What if someone went through all that headache and just provided all the instructions?!? 

What would you do with MSFragger on a high performance cluster? They processed 3,800 TIMSTOF files in 9 (nine) days! 

US HUPO 2023 Abstracts due in 2 weeks!

 

I know that the world ripped us all off and you probably also have a pile of concert tickets from 2020 from shows that might happen in 2023, and maybe you still have a voucher for a triathlon you were pumped about but your left knee, right ankle, and left elbow all stopped working recently enough that you're envisioning yourself drowning in the cold and unforgiving Chesepeake, but I'm sorry to say that 

YOU ONLY HAVE 2.5 WEEKS TO SUBMIT YOUR ABSTRACT FOR US HUPO 2023! 

Yes. 2023. 

Some amazing people talked me into being there! Join us at this beautiful event! 

You can register your awesome abstract here! 

I'm not going to slow down with these images and I hope it will get more extreme there. 

Why compromise on fragmentation types? OMNI-TRAP.

 I feel like the omnitrap has sort of a legendary pop culture sort of status. Were you at that one meeting 5 years ago when Dr. Zubarev talked about it? Did you hear whispers about it on the absolute wrong side of the Philadelphia Convention Center where my AirBnB was at ASMS? 

(That was probably about something altogether different...people in Philly tend to sound like they're mumbling...if you tried to get one of those crappy cheesewiz sandwiches from one of those places that are famous for them for some reason, you know what I'm talking about)

Well, guess what? This month's JASMS features THE DO IT ALL, FRAGMENT IT ALL, 

IT: 

CIDs (yawn)

it 

HCDs (....)

it
ECDs and UVPDs (oh?)

it 

EMLs (wtf is that?)

it 

SHCIDs (wtfingf is that?)

it 

IADs (now y'all are just making up acronyms)

it 

MS2/MS3/MS4s all using completely different fragmentation types, many I didn't list here because this article isn't the easiest to blog about for people with problems reading and writing letters in the correct order. 

Other than just showing off, why would anyone care about the 

Yeah, I learned out to make glitter text on the interwebs right now! I'm not going to waste it. 

The example presented in this study is working out a full intact protein sequence using all these different fragmentation types in tandem. That's cool enough to make glitter text about, right?

Machine learning for all the mass spec based -omics!?

 

(Gif above originally from S^%##y Robots which, although now something mostly defunct and replaced with something more grown up, really stands the test of time and is literally the first thing I think of when people say "learning machine" or "information artificialness" or ten other buzzwords.)

ON THAT TOPIC -- Check out this new pseudo special issue in JPR! 

It's not "pseudo" because it isn't special. It's "pseudo" because it is a collection pulling articles from across time and space around a single topic (though I think the oldest paper isn't even 2 years old). 

Summary:

FAIMS settings for ultralow (and ultra-ultra low flow) proteomics!

 

Do you have one of those big FAIMS things killing dewars of N2 in your lab...or...

...well...

...in either case, if your question is something like -- "Hey! What buttons would I press if I wanted to analyze <1 nanogram of peptide (why would you want to do that...?)?" And or "what FAIMS settings make sense for ultra low flow analysis?"

Check out this study where they worked it all out!

Study of DNA polyploidy cells is really a nice analysis of RNA-Proteomic correlations!

 

I started reading this because cancer polyploidy is really strange and very cool (honestly, I thought this was going to be background so I could better understand a recent paper out of Josh Elias's lab). Turns out the two things are only loosely related, and if I sort out the latter enough that my notes make sense, I'll come back to it later. 

If you aren't familiar with polyploidy, DNA divides in these cells, but the cells themselves don't. And some evidence supports them as being the origin of metastasic lesions. BIG DEAL. 

While I'm trying to force some new concepts into my tired old brain, there is a lot of this paper that is less difficult to grasp. 

Including some really exhaustive analysis of RNA to protein abundance correlations! No surprise, they aren't very good (the "correlations") but there is tons of great new stuff in this study, so the orange publish button will probably get hit. 

qPTM -- A nicely organized database of PTMs in some model organisms!

 

This wasn't the resource I keep dreaming of finding, but I do still like it. 

I'm not sure how this is better than PhosphoSitePlus and whatever the other thing is called that you can link to from UniProt now, except in terms of how it is organized. 

Think Mondays are a stupid day to read and just want to type something into a box? You can access this resource here. 

http://qptm.omicsbio.info/

How (Not) to Generate a Highly Predictive Biomarker Panel Using Machine Learning!

 OMG. I saw the abstract graphic for this recently

and I thought this was just a Tweetorial or something. It isn't, it's a real life super smart little review published recently in JPR.

100% recommended.