Sunday, October 17, 2021

Metabolomic links between depression and constipation in rats!


Science has been getting a lot of criticism recently, is the earth really round, does this SARS-CoV-2 thing actually exist, etc., What we need is a truly unifying study that everyone around the world can agree is something that needs answered and I'm excited to see that someone finally found the funding to explore: 

When a rat is constipated, is it depressed? When it is depressed, is it sometimes constipated? Does one cause the other? Is there a relationship at all? 

This team used NMR based metabolomics to explore these pivotal questions. 

To induce constipation, rats were fed white vinegar and activated carbon. It is important to note that this did not make them depressed, just constipated. 

For the depression model, the authors used a well-characterized system called the chronic unpredictable mild stress model. Just in case you aren't familiar here is a summary from the methods section of this study and the author's minor alterations: 

Which are, obviously, similar to the normal causes of depression in most mammals. 

What's the overlap? How do the two things relate to one another? Look, I'm not going to ruin this for you. You'll have to check it out for yourself here!

Saturday, October 16, 2021

Proteome Dynamics of bacterial sporulation!


Sporulating bacteria are responsible for all sorts of diseases. The little spores are dehydrated, surrounded by ridiculous numbers of layers of bacterial cell wall and other protective mechanisms that can make them resistant to extreme conditions. Some bacterial spores have been shown to resume vegetative growth after remaining dormant for thousands of years. Despite this, there is a lot we don't know about them.

Time to break out some new toys and do proteomics on them again, right? 

In this case, this group used a TIMSTOF Pro and 55 minute gradient, which probably seems like overkill for a bacillus that only has 3,000 genes, but hey, it's just an hour per run! 

What did they get for this? Well, this paper is extremely short and just kind of shows that proteomics can support 50 years of molecular biology used to figure out this model organism (B. subtilis), but all the files are on ProteomeXchange here

Thursday, October 14, 2021

Very different proteomic response to Vitamin C in male and female mice!


This new study ASAP at ACS's Proteomics journal throws me because of how very different these profiles are. 

Obviously, we need to be thinking about the effects that gender and genetics have (related point out of our lab)...

...and we all know this, but do we really know it? 

This group did expert level discovery serum proteomics on an Orbitrap Fusion, I forget which one now, and then they did expert level PRMs (maybe because they also couldn't believe how different these results were)? They spiked in the Biognosys iRTs for controls and I can't find a fault with their methods or data at first glance, and the serum proteomic response to something I consider as benign as Vitamin frickin' C in their models is very different between male and female mice. This really drives home for me how much we've got to think harder about proteomics in the context of biology and expand our matrices to get it all in. 

Wednesday, October 13, 2021

The mystery behind a 1,000 year old paintjob solved with proteomics!

Oh. This is too cool, and not the paper I meant to blog about this morning. It clearly says "top down thio-TMT one" on the sticky note on this monitor, but I found this while looking for it. (Top down TMT is super smart, though, and maybe I'll get to it) 

How do you paint something so that it stays red for 1,000 years? You start with mercury sulfide and someone solved that mystery 30 years ago, but there was something organic that held it all together and that was a mystery. 

Enter proteomics! With a tiny tiny tiny scraping of it and an Orbitrap Elite system this group shows that it was probably chicken egg and human blood. 

My favorite/least favorite part of this awesome study? 

Archaeologists have lots more things they want to run proteomics on, but it's too expensive??? For one, you can get an Orbitrap Elite like this one for $60,000 (often with a warranty) from ReUzeIt or ConquerSci or other second party vendors. 

However, if you're an archaeologist who has something awesome to do proteomics on I guarantee you we can find someone out there in the proteomics world who has time open on an instrument to squeeze something in. That goes for most things. Even core lab instruments have downtime here and there on Sunday nights. Network, yo! 

Tuesday, October 12, 2021

The N-Glycoproteome of K562!

 As you are dutifully moving away from the last experiment you'll ever do with HeLa cells or protein digests and starting anew with superior standards that are not unethical to use or sell, my first recommendation is the Promega K562 digest. (I 100% side with the family, and that company was warned this would happen.)

I've spoken to Promega about how they generate it and how they test for these scary mycoplasma things (thoroughly) and after running it at least once every day for QA on our TIMSTOF with extremely reproducible results (our daily QA is: 30 min LC gradient, 15 cm PepSep Reprosil C-18 1.7um with 3cm trap of same, "if 200 ng of K562 is less than 3,300 protein groups on FragPipe QC workflow, stop, something is wrong", which is altogether kind of insane, but when the system is freshly clean we come close to 4,000)

If you're thinking: "wait, is it as complex as that other cell line?" Check this out!  

5,000 N-linked glycopeptides?! That should be enough to QC your glycoproteomics workflow! 

As another option, NIST developed a liver homogenate lysate for proteomics. I don't think it's shipping yet, but it should go live soon. 

Monday, October 11, 2021

Great Clinical Proteomics Talk this Wednesday!


Interested in getting these mass spec toys into the clinic to help patients? 

Chris Shurford is this year's winner of the Michael S. Bereman Award for Innovative Clinical Proteomics. He's giving a talk on his work this Wednesday.

Sunday, October 10, 2021

Proteomes ARE proteoforms!


In all the excitement of "next gen proteomics" (i.e., the technologies that will soon make peptide quan by mass spec seem like a great way to waste both money and time while getting largely inferior data), mass spectrometrists will have several avenues to explore before those that won't adapt in time will end up relocating their labs to homeless shelters. 

This review in Proteomes discussed one such avenue nicely provided by evolution: 

Proteoforms! The basic fact that protein quan, while often better than trascript quan, is largely kind of dumb from a biological stand point.

Very few systems (at least in humans) are actually governed by a cell that made a whole bunch more of a protein. Sure, some of them do, but really that's not how any of these systems really work.

It's more like: kinase A phosphorylates target B and C. Target B translocates to the ER where it splices protein D. Phosphorylated targed C and spliced protein D form a tetramer with 2 of each of them and that tetramer migrates to the membrane and opens a pore. Most of the time 10 other things are ubiquitylated/ubiquitinated/sumoylated and THOSE are the actual phenotypes! 

When that new guy shows up at your university with that benchtop thing that can quantify 7,000 proteins in 4,000 samples before lunch, that's your survival plan. Move toward quantitative PTMs and start saying the word "proteoform" all the time, cause that's the currency we need to be using. 

I really like the paper above

I also think this is worth a read

Saturday, October 9, 2021

Single cell proteomics (and preprints) take center stage!


Chances are you've probably seen this picture of Ying Zhu and this feature in Nature a couple of weeks ago. If you haven't it is a solid read. Obviously, I'm biased, because basically all the plans I have for something called "tenor" involve applying single cell proteomics to better understand how drugs work. 

While this is a great read and highlights papers that we've talked about a lot in lab (with the exception of yesterday's ("yesterday"'s) post which I missed somehow. 

The thing that really jumps out to me is this -- can you imagine even 3 years ago, those old fogies (is that a word?) at Nature running a feature on 15 studies (whoops, they added one they missed) 16, studies where 7 of them were preprints? 

I think this is really important because preprints are definitely at a crossroads. 

The data is coming in and it's pretty conclusive: preprints are not getting cited in peer reviewed literature. Now, if a preprint is there and it gets accepted, it appears that having preprinted the study does help. Honestly, I think this may just be the fact that there are two places where the study will pop up in a Google search.  (Here is one reference, caution, this is the direct .pdf download). 

This is despite a completely scientific and unbiased survey that I performed in the Spring that came back with these results. 192 scientists is a pretty big number. There are only like 8 million in the world. 

A grant I applied for recently didn't consider preprints at all. They simply couldn't be uploaded or linked. I'm new at this academic thing and don't know the rules yet, but I'm going to go ahead and suspect that wasn't an isolated system, so I'm going the self preservation route and keeping my best data secret until it gets accepted somewhere for now. I gotta get some grant money at some point or pack my desk and I have a fat baby to feed now. ]

Again, we're at some sort of a crossroad. What's best for the world? Obviously preprinting data, open sharing all of it and pushing to make the world a better place. What's good for individuals? Hopefully it will be those same things! This Nature feature suggests that maybe it can be, but I'll argue we ain't there yet, hon. 

Friday, October 8, 2021

Multiplexed DIA!!


This is from April?!?! How on earth did I miss this? 

Tiny window DIA on multiplexed samples!! My first thought is "of course that would work, but aren't instruments too slow?" 

I guess not, because this data (HF-X) looks pretty great. This does explain some questions I had regarding another preprint from this group where I was wondering if I missed something. Guess I did! 

Thursday, October 7, 2021

Is it really acetylated and where you think? Check the chromatography!


Acetylations can occur on multiple amino acids in your peptides and it's no secret that most search tools struggle with localization of modification sites (that's why we have A-score and ptmRS to help).

This group looked at loads of modified and acetylated peptides in terms of how they elute on reverse phase and other types of chromatography

If you're really stuck on that site and whether it's: A) real and B) where it is located, this might help? 

Wednesday, October 6, 2021

Immunopeptidoproteotranscriptogenomics illuminates the immunopeptidoproteotranscriptogenome!


This is a nice walkthrough on using transcriptomics to guide the discovery of new neo-antigens, including the use of synthetic peptides and MHC binding tools to help support your results. 

You might get to the flow chart on how they did the informatics and think about pursuing a different field of study, because....ouch....there are 40+ steps here. It sure seems simpler to not do the transcriptomics at all. I wonder when we'll hit that tipping point? Some journals don't require that you validate your mass spec data with western blots or other rabbit blood based techniques now. How far are we from just using the mass spectra and not needing the noisy crazy looking next gen sequencing data? 

In terms of the details, the samples were a leukemia monocyte cell line and the data was generated on an Orbitrap Fusion in high/high mode (orbi-orbi) with HCD. Most of the data analysis on the proteomics side was done in PEAKs. 

Tuesday, October 5, 2021

What's in a name -- picking reviewers, by Dr. Eyers!


Link here.

This is short and inciteful for anyone who, perhaps, just applied for their first ever academic grant and found during the process (at least through the system they used) you can't list preprinted work or even mention it, so maybe you look like an idiot for applying for something you can't list any proof that you can do. If you've been locked in your office working on getting papers accepted in journals so you look like less of an idiot when you ultimately reapply, good advice is nice to see and receive! 

Friday, September 17, 2021

AlphaTIMS -- If you have a TIMSTOF you'll use this all the time!


Getting a good snapshot of this great tool was tough (and file name dedacted just in case it matters to my collaborator).

TIMSTOF files can be gigantic and digging through them for peptides that you know are there can take a really long time. Apparently some people at Max Plank noticed this fact and decided to take a swing at fixing that part with! 

Have you done enough reading for this lifetime? You can just download AlphaTIMS at this Github

Lots of folders? 

There is a "One Click GUI" and the software is pretty straight forward. You can filter your data by scan number, retention time, quadrupole settings, 1/k0 value, ion intensity, etc.,

What you'll find is that it's really easy and extremely fast to hunt down specific ions in your data. I'm still a little unclear the differences between the m/z index isolation and quad isolation functions. Wait. Nope. I just needed to type this (and maybe espresso to do espresso things) for it to click in my head. The quad filter is for the literal quadrupole. The m/z index is just my m/z view window. That's why the former doesn't do anything if I don't have fragment ions selected. 

You can also use this to kick out data as HDF, CSV or MGF. 

My understanding is that this is just the beginning of a whole bunch of new tools for this system that are in various levels of development, but I'm using this one constantly now and just looking forward to seeing what is next. 

Wednesday, September 15, 2021

SureQuant-IsoMHC - absolute quantification of antigens in patient samples!


MHC/HLA peptide studies for global proteomics generally start with "we cultured approximately this many cells.... generate enough peptide for a single LCMS run" 

This works for pure discovery and it still seems like we need to get the discovery aspect right, but for patients we need to crank up the sensitivity and here is a really impressive new way of doing that

SureQuant uses internal standard heavy labeled peptides to trigger focused targeted analysis. I haven't used it myself, but I also hadn't seen a reason that screamed to me that it was a necessary tool until now. High resolution absolute quan of HLA peptides....from a punch biopsy?!?!?  I strongly recommend checking this out. 

Tuesday, September 14, 2021

Comparative solutions for genotyping of human hair with LCMS!


Ruhroh, Reorge. Glendon Parker is still out there trying to rain on our free proteomic data parade in this new JPR paper

We've been doing some classical proteogenomics and getting the "publicly deposited" genomic data for it was a huge hassle. Justification forms and waiting months and we were under a really strict deadline for the project that was imposed by the people who had the genomics data. Longest part of the project? Waiting to get approval for the whole exome sequencing data.

You have to make sure that someone isn't going to use that 250 GB RNASeq data file to extract personally identifiable data from the patient for nefarious purposes. The 24 high pH offline fractionated normal and tumor data from that same patient? Pull that down as fast as your internet connection will allow. 

Could you identify that patient by the single amino acid variants you can find the proteomics data? 

....let's umm.....go with...wait! change the subject!  (Dr. Parker, that's enough from you and your group. I don't want to wait 2 months to download every .RAW file from every preprint. I'll forget to do it!) 

Look, we are going to have to tackle this at some point. Either the genomics people are being crazy paranoid about personal data, or we're being lackadaisical. 

I'm definitely being that (did I spell that right or is the spell check off?) last word, because I sat in on a webinar the other day and someone showed a slide that I recognized as a list of single amino acid variants that you can see in my personal plasma that are confirmed by my personal whole genome sequencing data.  I've got some plasma proteins that look downregulated vs pool in some analyses because of the variants. I use the slide to point out issues with extreme ratio quan in some LCMS tools and why we need to think about variants and I appear to have shared that deck a lot. 

In a world where the terms "pre-existing condition" and "life destroying medical bills" (still the #1 cause of bankruptcy in my country where people are dying at a really depressing rate because they are using expired or black market INSULIN because they can't afford the real stuff) maybe we wouldn't actually care about what of our personal data is out there in the world. 

But if we actually care, we might want to actually care about all of it. Not just the stuff that you need access to an HPC cluster to properly process. 

If you do want to identify someone by their single amino acid variants IN THEIR HAIR, this new JPR study will tell you which hardware solution to use for it and how to best set it up. 

Monday, September 13, 2021

Inactivating coronaviruses!

Remember that COVID thing in the spring of 2020 that was such a pain in the ass? Holy cow, how many of us needed THIS paper? 

Thankfully, it turns out that those things aren't all that hard to deactivate, but what a great reference to have in case a coronavirus becomes a problem again!

Saturday, September 11, 2021

Scale up (or go to lab less!) with 384 well plates on your LC!

You probably already knew this, but until I didn't until recently. Every EasyNLC that I've ever used has been set up with 6x8 autosampler vials by the previous owner. I guess I've never had a new one, which... come to think of it... might have something to do with why I've been so mean about them, they're pretty finicky about being abused.... oh well... well, if you didn't know you can set them up with plates!  

Even if you use long gradients I bet there have been times where you've made tough decisions with 6x8 about either going in over the holidays or skipping some things because 48 x 3 hours still isn't all that long. We've been using 96 well plates, but with 30 minute gradients I've made hard decisions more than once. 

ENTER THE 384 well plate! 

As always don't take my word for any of this, where did you get this information? The internet? 

This is what I'm using

384 well plates (there are less expensive options, but there is some weird virus thing still disrupting stuff, some other 384 well plates that I've ordered for sample prep have been backordered for 5 weeks)

Webseal tape

Your EasyNLC probably already has an option for 16x24 plates. Woo! You're done! 

(I'm joking! Stop! You'll break your needle! I was going to put this below the picture but had a fit of conscience.) 

My A1 position was around B2.4. I don't know if I could have ran it without calibrating first, but if I had, my needle would have pushed right into the space between B2 and B3. 

You'll need this manual if you have an EasyNLC1200 (WARNING. DIRECT PDF LINK.)

You'll also want a flashlight (or a fancy phone with a flashlight. I made my Siri British somehow and I have to say "Siri, turn on the torch". In the basement here it says "working on it" then "working on it" then "something went wrong" in a snarky Alex Kingston need a flashlight.

As shown in the top picture, you need to remove the steel plate on the left side of your LC. The 3 screws are snap screws. Turn them counter clockwise 90 degrees and they'll pop free and the plate will come off. 

Install your 384 well plate (without tape on it, unsealed!) and start the calibrate procedure (the one in the middle of the screen)

The calibration isn't a complex one like an OpenTrons you just need to position 2 points. The bottom of your A1 well and the bottom of the well at the complete opposite corner of the plate.

If you touch the A1 position on the screen the needle will go somewhere near there, now you need to manually locate the bottom of the A1 well. 

You can see the A1 well perfectly so that one is easy. Use the extremely European touch screen controls (needle in/out is not into your well, it's into your instrument. you'll see what I mean) to navigate your needle to above the A1 and down until it touches. 

Now -- Take a picture of your needle coordinates. My A1 was this, I think, but what I care about is the depth of the needle in A1 because you can't see into the far opposite well (I don't know what that one is called yet, I only filled my first plate to N6) 

NOW you can capture your coordinate! Hit the green well above the opposite A1 and repeat the process. Get the flashlight. If you have bifocals, get those too! 

Without having the picture of the depth of A1 this will take forever because the plate will move just a little and you'll freak out that you hit bottom and will be afraid to go further. I verified that the bottom was just about perfectly the same on both sides by raising the needle 0.1 arbitrary units and finding just a little wiggle room in the plate). 

Now, I'll probably keep using this tape that I linked above, there is a very satisfying "POP" sound when the needle goes through it, but it does leave a permanent hole in the top, so you probably can't inject more than once out of a well unless you've filled a blank. Anything you worry about evaporating will probably evaporate, other than that, I'm up to L and the QA sample looks legit. Everything is working.

Oh crap. Almost forgot. 

You'll need to adjust your plate in your instrument configuration so it knows you did this. On a Thermo instrument it's the little double gear thing probably (which is now an App in Windows 10. I hope it makes this 1990s program feel totally hip or something).

On the Brukers it is one of the circled things. I always try them both first. 

You'll need to just let the queue know that you've got a 16x24 installed! Now you're done! 

Friday, September 10, 2021

Carrier channel math on Orbitraps! Something proteomics people agree on!

There is so much important foundational work being done in single cell proteomics right now! I know several people who are very interested in starting to do this once "it's all done" and this is the stuff we need to get through. 

1) has some of the prettiest plots I've ever seen for proteomics data 
2) started in a completely different and really smart way to work out carrier channel ratio compression (using swapping TMT126 chanel for TMT0 to get some sense of ground truth.)

And what's really impressive is how consistent every study that has looked at using amplification via (SCoPE-MS or BOOST or whatever terminology this all lands on in the end) has been in the end. Now, this is the most conservative estimate to date, but everything has been pretty much in the same ballpark. There is a single order of magnitude between the lowest estimate and the highest, we're somewhere between 20 and 200).

Always worth considering here, and I apologize if I'm starting to sounds like a broken record (whatever that means) but every study so far has been on an Orbitrap, with only one group using the bigger D30 Orbitrap and everything else using the D20. Not to oversimplify the physics involved, but it's not just coincidental that the larger Orbitrap measurements have biased toward higher carrier. 

We've known all along that while the Orbitrap is one of  the single most awesome things that has ever been invented, the instrascan linear dynamic range is NOT why you buy one. You want big instrascan dynamic range that's linear? Get a QQQ. Everything else falls in the middle. 

Back to the paper: 

Another really good paper I'll be referencing all the time has done the carrier ratio suppression math and just to flag why I'm so impressed by them getting to the same conclusions: 

We're a 20+ year old field and not one single one of us can digest a sample the same way. Proof in point? 

Paper earlier this year? 

1:50 trypsin with overnight digestion at room temperature

10mM DTT for 30 min at 56C

55mM IAA for 45 min at RT

Some other changes: 

In Claudia's new paper we see something slightly different (surprise)

1:100 trypsin overnight at 37C 

50mM DTT for 30 min at 37C

40mM IAA at room temp for 30 min

Does this change anything? Probably not, but if 2 proteomics people from different labs land on the same conclusions about anything it's really impressive. 4 or 5 studies coming up with very similar conclusions? Maybe this current generation of proteomics trainees will fix the rest of it! (I'm super pumped they both used the same alkylating agent.) 

Thursday, September 9, 2021

Bispecific antibodies -- How to ruin a cancer cell's day with a neo-antigen!


I guess this isn't new news, but it's news to me! Hard copies of Science kind of arbitrarily end up on different coffee tables and I don't always get to flip through in order. This overview of the field and the promise it has is from March is a solid and approachable read

Importantly for me, it helped put this other paper I've been thinking about here and there into context and suddenly -- it all clicked. (Don't you love when those disparate neurons realize they're all related and it all falls into place?) 

You know how we're all called on from time to time to identify those awful MHC, HLA, endogenous peptide, surface neo-antigen things? If you have you've probably been at a point where you've struggled through it all and handed over a list to someone and:

A) You never hear from them again and you might wonder what they are going to do with it


B) (Way way worse) they ask you something like "okay, which one should I spend a lot of time, money  and resources validating?"

In this study you see behind the scenes and get a glimpse of both answers! 

This group prioritizes the neo-antigen targets through a process called MANA-MRM! 

Then they build a bi-specific antibody to the neo-antigens and totally f' up some cancer cells with it! 

As much fun as it is to bicker about trypsin concentrations and p-values, I will admit that I get a lot of inspiration from the couple papers a year that show LCMS can address biological or medical challenges. 😇

Wednesday, September 8, 2021

Isotopic Distribution Calibration!


Holy cow. I'm just going to leave this here until I have time to come back to this. If you're solely doing shotgun proteomics by LCMS this might not be the most interesting thing you've ever seen.

If you're from time to time digging all the way into your data to determine the difference between a PNGase cleaved glycopeptide or a citrullination and a C13 isotope or 4, or you've wondered things like:

 "why can't I use isotopic distributions at 700,000 resolution to determine if this is really pomegranate juice because of the different carbon fixation mechanisms used by different plants"  (hint, it still isn't pomegranate juice, global production is still only a small fraction of global consumption)

 or attempted to use isotope shifts to determine how old the material is in a drilled core geological sample is

or tried to distinguish between natural and synthetic compounds 

You've probably been extremely annoyed about the inability to calibrate your isotopic distribution and maybe thought of crazy ways to do it yourself and then given up. 

If any of this applies to you: