An oddly satisfying reviewish paper on MultiOmics

MultiOmics might be my favorite word right now. I'm not the only one who is digging it. Lots of papers seem to have this new word in the title right now. And with good reason!  We've got so many ways of globally profiling what is in the cell that the next trick is going to be integrating all of them!

For an oddly satisfying review of the topic, you should check out this open access study in Nature Communications.  In the blog title, I'm going to leave the term "oddly satisfying." I'm not sure what it is -- is it the way this article rapidly tears through all sorts of -omics things? Is it the nice bold lines used for the illustrations that remind me of an Archer cartoon? Is it the cold medicine that has been slowing my brain down? I'm not entirely sure. What am I writing?

What I do know for certain is that, while this is an original study on the multiomics of E. coli it reads a whole lot like a review, cause it knows it has to introduce the reader to at least one -- if not many, of the things they're looking at. It also reads that way because this is almost entirely a meta-analysis of data from repositories. They do appear to do some of their own transcriptomics. Cause, you know, nobody's ever done RNASeq on E. coli before (though, more likely because it was easier to run their own samples than to download someone else's 2TB or RAW data).

In the end, though, the best part is that it shows how all these technologies are complementary. They can detect problems in their bacterial test system from data at the protein and protein structural level as well as in the fluxomics and at some of the transcriptional stuff. Key points here -- sure you get extra info from all these other techniques -- but it doesn't mean you have to do them to get to your answer!  There are many ways to get to the same answers!

Ever tried short tryptic digestions and failed? Try Partial FASP!

This brand new paper in Nature Chemical Biology  probably isn't there cause its got a really cool (and reproducible!!!) method for partial tryptic digestion. It probably helped, but it got there cause it conclusively identifies a new post-translational modification that can occur in histones and that has all sorts of sweet biological relevance.

But I'm a methods nerd and this is something I've tried to do -- lots of people I know have tried to do -- and I've never seen it work reproducibly well!

And theirs works!!!

They had to do it cause the PTM they are looking for falls into a domain where the trypsin would cut it up. And the method is really straight forward (and FAST!) and it makes peptides that shouldn't be ETD compatible --- big and ETD compatible!!!

I'll probably write more about this paper, cause I'm fascinated by the PTM, but its Halloween Friday. Check out this great paper!

In vivo protein structural information!?!?!?

We had an AWESOME meeting in Bethesda yesterday. A highlight of my weird job is that I have input into picking speakers for some meetings. The downside is that with 2 rooms running I couldn't see every talk.

This is one thing that blew my little mind AND I know I can talk about it cause it was recently published in ACS here!

I've got a minor obsession with what you can do with mass spectrometry and fast photo-oxidation of proteins (FPOP). 1) Cause its super cool and 2) I don't know how to run an NMR. But...in the end you can probably learn more about an in vitro protein or protein mixture from sending it to the weirdos who run the NMRs down the hall.

But...this...study blows the doors off of the little my brain has stored on protein 3D structures and interactions. This group shows that they can separate cells -- individually -- and then FPOP them -- and get information on the inside and outside features of the proteins --- from their in-cell environment!!

WHAT?!?!  I know!

This is a method development paper, obviously, but think of the implications here -- if you could get enough coverage -- you could have info on how your mutation affects the protein structure and even the protein-protein interactions of any of the proteins in that cell!  There are obviously some challenges -- huge search space and MS/MS sensitivity among others...but WOW! Right?!?!

Widespread K methylation in plasmodium -- and other things!

My Destroyer is getting set to reanalyze some old data while I'm out of the office today, thanks to this new paper.

In this, they do some really intense antibody-based pull-downs specifically looking for lysine methylations. And find them...all over...in Plasmodium falciparum. I'm always hunting for explanations for the spectra that we can never seem to explain from these organisms, and maybe some of them are thanks to methylations all over the place.  My first thought is throw in lysine methylation and just see -- then my next obvious thought is...is this inhibiting trypsin...?

So...in trying to hunt down some info on the topic...I find this paper...

...that strongly implies that maybe I ought to be doing meta-analyses on other proteomics datasets cause lysine methylations may be involved in all sorts of things outside of where we expect them to be (histones!)

One of the things that comes out of this Tuesday morning post-espresso paper binge is this fact: lysine methylation has been predicted the 4th most common PTM in nature? What?  Who decided that?!?

Turns out these guys did back in 2011!!

Check out this chart!! (Click to expand if you need to)

They looked at how often we observe PTMs compared to how often PTMs ought to occur. We know we're Phospho-biased, but it popping up as number 2 makes it seem like not such a bad bias. (I wonder how this chart looks now, considering how the glycoproteomics field has blown up the last few years?)

Does this post make any sense?  I know it lost linearity somewhere. What the frack is a geranylation?!?!?  I'd better hit the "Publish" button before this gets any more chaotic!

The phospho kiss of death!

I stole the idea for this post title from someone else, but its totally fitting for both the paper and for Halloween week!

You know how we normally figure out how a mechanism works in something really simple -- like a bacteria, and then we figure out how more complex organisms like us do things? This new paper in Nature from D.B. Trentini et al., shows us 2 super cool things:

1) Sometimes we figure out how the more complex organisms work first and
2) What the heck Arginine phosphorylation does!!  (And this is why it is in Nature!)

We have to break down and get rid of old proteins. Or our cells will all just end up completely packed full of old proteins (sub-ideal). In us, that is one of the things that ubiquitination does. It marks old proteins for degradation so we can get them out of there.

Until now, there wasn't a clear understanding of how this works in prokaryotes. Turns out...it is mediated by arginine phosphorylation (at least in the awesome model gram positive Bacillus subtilis!)

They work out here that a protease called Clp (fitting?) selectively finds proteins with Arginine phosphorylation sites and eats them up. They find it by doing Clp pull-downs of normal and heat shocked cultures of the bacteria and then back it up by elegant in vitro assays as well as with knockout strains of bacteria. You couldn't be more thorough.

Interesting note here for us lab rats --

--can I actually post this? (Nature, please don't sue me, if this is a problem please see my contact info under disclaimers and I'll take it down!)

What is highlighted above -- I don't know how to do, but I think its awesome!  Now that there are nanoflow UV detectors, does this mean I can rapidly determine during my runs my approximate tryptic digestion efficiency??  This bears further examination.

The phosphoproteomics is done with CID/ETD on an LTQ Orbitrap Velos Pro. (Unless there are 10 of them at IMP, this might be coming up on being the most famous Velos Pro in the world -- what a work horse!)

Anyway...super solid paper that absolutely deserves to be where it was published!!!

Mass spectrum jewelry!

This shop makes 3D printed mass spectrum bracelets and handmade metal spectrum pendants! What?!?! I know!

Opening Proteome Discoverer 1.4 Results in PD 2.1

Great question from a reader! Figured it deserved its own complete post while I'm in-between meetings!

You've got all those awesome MSF files you've processed in PD 1.4!  How do you open them in PD 2.1?  You definitely don't need to reprocess them all!!

Do this!

1) Make a new folder

2) Don't add any RAW files or templates or anything. Just blank on all that stuff

3) Where you add your RAW files -- add the PD 1.4 MSF files!

PD 2.1 should recognize all the stuff. Here it recognized that these 2 processed TMT fractions are -- TMT fractions and that I used the TMT 10plex. Now...I do have the RAW files in the same place where these .MSF files are. I'm not 100% sure if this is essential or not. Its just how my storage drive is organized!

Go to your analysis results and they'll be there as well!  w00t!

Now is the conversion part!  Highlight and Reprocess your data as a new MultiConsensus report!

Then make a Consensus workflow. I just chose one of the common default workflows.

Run it!!!!

TAAAADDAAA!!!

MS3 TMT -- in the ion trap or the Orbitrap?

Orbitrap Fusion devices have the well-advertised ability to do the SPS MS3 technique that was originally developed by reprogramming an LTQ-Orbitrap Elite.

In the typical technique, the ions flagged as tagged peptides are fragmented in the ion trap and then large precursor fragment ions that contain the tagged side of the peptide are isolated for HCD Orbitrap scans.  At this point many studies have shown that this technique improves quantification (especially "ratio compression" problems), but doing 2 high resolution (typically 120k and 60k) Orbitrap scans isn't the fastest way to run your system.

Jane M Liu et al., think that one way to speed it up is to do the MS3 in the ion trap instead!  In this study they take some yeast digest and go to town, comparing the maximum number of quantified peptides with the traditional scan vs MS3 ion trap.  The figure above sums it up pretty darned well.

Now....you'll notice they used the TMT 6-plex. Cause the 20mmu difference between the neutron mass discrepancies in the TMT 10-plex can't be determined when using the ion trap, but its an interesting idea for experiments that are using 6-plex for sure!!

One other lab has already reproduced this!

Two quick notes: 1) I LOVE Twitter! Real time tests on study reproducibility?!?! The future is now! and 2) I thought this Lumos was just delivered.....UCD doesn't mess around!

A less invasive proteomic test for pancreatic cancer!

Pancreatic cancer still sucks. Its still killing people indiscriminately from all parts of society -- including Munsters...  Every talk and paper I see on it says that early detection is the key to beating it. Problem is, the stupid pancreas is tiny and checking it is super invasive (long needles....).

Jana M. Rocker et al., says there is a way less invasive way of checking out what is going on in the pancreas!  In this study these authors compare the proteomes of patients who have the pancreatic fluid extracted to fluid obtained through by adding an extra step to a routine colonoscopy. Turns out that the fluid (called whole-gut lavage fluid).....

...contains a whole lot of the proteins that are monitored when you stick a needle into someone's pancreatic duct!!!

Awesome, right?  Lets move this sucker into the clinic! This method is (IMHO) ready for the clinic. The sample acquisition, sample handling and prep, instrument method, everything is clearly laid out so your local gastroenterologist can add checking your pancreas proteins (!!!) to that panel of tests they're going to do while they're already in there.

More early detection, and less cool people being blindsided and taken out by pancreatic cancer!

P.S. I'm not the only one totally digging this method. This Nature editorial highlights this paper! 

Informed search space and value of controls in glyoproteomics!

Bioinformatic limitations are seriously affecting all of us  glycoproteomics research. There are just SO many possibilities. Multiple glycosylations sites, multiple glycosylation patterns per site, numerous isomeric glycans, etc., etc.,

This new study shows how big the problem is -- and provides a way to accurately assess how your search parameters are affecting your results!

They do some glycoproteomics on samples of different complexity while spiking in the same protein, AGP1.

A quick Google Image search says....this is AGP-1...thats a bunch of glycans!  Now that they have a good control glycoprotein, they can see how changing the search parameters affect their ability to detect the glycopeptides they know are there.

And the results are....scary. In the most complex samples if they just do a "naive" search where the search engine is looking for any possible glycosylation sites or combinations....the likelihood that they can find their control glycopeptides diminishes to almost zero.

Take away from this study?
1) More specific (informed) searches are going to be a whole lot better for more global studies AND
2) Having a good control like this AGP-1 protein is a great way to make sure you're on the right track!!

MAP ALL THE CELLS!!

This initiative is getting a ton of press and raises some interesting questions (in my head, at least!). Depending on who you ask and where you draw the lines of "what is a different cell" there are something between hundreds and thousands(!!!) of different cell types in a human body (lets ignore our microbiomes where we're probably looking at orders of magnitude more, LOL!)

The idea is that a bunch of smart genomics researchers need to make some lines -- "this is this cell" and so on -- and then map the crap out of them!

The interesting question in my mind then is this --- if we had maps like this -- should we be using them in proteomics? We know the more specific the database is the better our results turn out -- so, for example, if I am doing an analysis of PTMs on glial cells -- should I be using a FASTA generated only from glial cells?

Searches would be faster and you'd think the FDR would be better(?), but you are placing a lot of trust in the people who originally isolated those glial cells in both the genome and your study as well as in the FASTA annotations. I can't wait until I have the option, though! Get sequencing!

Personal note

Wow! This blog has totally sucked more than normal. I had some personal stuff to take care of, like marrying my dream girl!!!  Shoutout to Match.com for allowing some oddly specific search terms so I could find her -- and to Dr. Norris for officially joining me on this life adventure!!

Panorama Auto QC (SuperProCop!) -- QC your MS system on the Cloud!

I've possibly rambled about my love for SProCoP on this crazy stream-of-consciousness that is this blog. If you haven't ran across these ramblings, SProCoP is a quality control/assurance system that can be added into Skyline. You can get it here. And you can read about it here.

If you've caught any talks recently from the Skyline folks, you've probably heard about the awesome projects -- like Panorama and Chorus which are utilizing Cloud resources (which is probably gonna be super cool for most us us -- except you guys at government/military facilities that probably block this blog already...I knew China and Thermo Scientific had blocked this thing for all their employees-- but now even the U.S. Army? Come on!)...wait, was that a complete sentence?  Nope!

This new study shows that we can combine these things -- SProCoP and the processing power of the Panorama cloud into -- SuperProCop! (My name) Or... Panorama AutoQC!  This is how it works...

So...if you are one of those cutting edge targeted labs that is setting up assays in Skyline and processing/sharing results via Panorama -- BOOM!  Quality control tools at your finger tips!

Okay, so there are lots of open source tools out there that will "QC" your system. And some of them are awesome. Maybe some of them are awesome if you feel like writing a lot of source code and Python. This one? Totally awesome.

Check this thing out!

What is it? Its a chart I just generated showing 2 peptides that the Beremen lab uses at NC State uses and their relative retention times over those 2 YEARS(!!!) of runs. Look at the bottom chart, sometime around the end of January this year, the retention times of both peptides started drifting kind of high. Wonder what happened there?

Well, what if I go and plot the FWHM for the peptides picked up from each peak by flipping the Panorama AutoQC to plot the super cryptic function "FWHM"?  And then I forget how the highlighter function works in the Microsoft Snipping tool and draw a line through the peak I'm indicating and then a HUGE CIRCLE? (And I'm on vacation and way too lazy to fix it?)

You'd get this!

The first peptide got super wide at that point! The second peptide got crazy narrow. If you knew something about these peptides, then this could indicate what went wrong in your system. More importantly (!!) variations from the norm could indicate that something is wrong before you collected sub-optimal data!

Time to shake up the clinical proteomics paradigm?!?

This paper is from March. Not sure how I missed it, but I absolutely love it!  Its from Philipp E. Geyer et al., and comes out of some lab at Max Planck.

The gist here is this -- maybe we've had this whole clinical plasma proteomics thing a little backward. The way we do it:
1) Get the deepest possible plasma proteomics profile you possibly can on a few people with your disease.
2) Repeat on people without the disease. Heck, maybe add them together and do Super-SILAC or TMT or something. Either way -- spend a TON of time on just a few samples
3) Find something cool -- probably with bad statistics, cause you had an N = 2 or 10
4) Do a targeted analysis that is much more rapid on a bunch of people.

This paper isn't really revolutionary or anything, but it does shuffle things up a little, and I like the logic here. They work out an extremely rapid method starting with a pinprick of blood. Okay, actually, I am going to change the picture above and put in the summary picture cause its better at explaining it -- TADAA!  What they did is in the picture above!

The highlights here -- rapid, reproducible sample prep followed by EXTREMELY rapid nanoLC separation and analysis on a QE HF and then label free quan (with cumulative analysis at the MS1 level -- i.e., if a peptide retention time and isotopic profile matches within a tight parameter, the peptide doesn't necessarily need to be fragmented for MS/MS in every single clinical sample -- this is going to be a repeating theme coming up!)

Short gradient? Yeah! 15 minutes for the main elution stage!  Out of their cohort they identify and quantify 345 proteins.

345 proteins? That's terrible, right?!?  But this isn't the focus. The focus here is getting rapid reproducible reliable realistic confident(!!) proteomics on as many clinical samples as possible. Out of these they come up with 50 or so of the approved FDA clinical biomarkers. 50!! (or 49!)  For comparison sake, if they do some fractionation and run more traditional 100 minute gradients they only come up with about 15 more. Waaaaay more time and not that much yield.  What do you want out of those patient samples? Here they're saying -- 350 protein quantified in the time it would take you do ELISA for a couple. Rapid diagnostics and huge cohorts real fast!

This paper is really written with clinicians in mind. It talks about blood draw techniques and how they can determine (from the proteomics) whether it was not done correctly (excessive lysis)

I do have one criticism. And I hate to say it after such a nice study, but.....these authors go to an awful lot of work to build a super awesome canned method. If I was tasked with setting up a clinical proteomics diagnostics lab somewhere -- here is the complete template -- I'd put copies of this paper up on the wall everywhere and follow it to the letter. With the exception of the chromatography.... The authors use a custom prepared in house 40cm column. And from the CVs they get from the study and everything it is obvious they are really good at it. But I'm not. Want a really crappy nanoLC column prepared? Give me 3 hours and access to your bomb thing. I'll screw up a 10cm 15 out of 16 times. 40cm? Not a chance. I don't think I'm the only one that would have trouble making a good reproducible 40cm 75um column.  If this paper had used a commercially available QC'ed column from any manufacturer I think that it would be just a little bit better for being a canned template -- and I wouldn't even look -- I'd just order 100 of whatever they said from whoever they said to get it from and be running samples a week later.

Outside of that? Seriously, just a killer paper. Maybe the paradigm is wrong -- or it isn't the best solution for every study, whatever -- but this is the best study I've seen with a solid alternative.

Side note: No plasma protein depletions!! WoooHoooo!!!

Proteomics proves Neanderthals made jewelry!

Need a way to put that cousin you have that left his/her stable job at the accounting firm to find him/herself to make surprisingly complex silver wire jewelry into some sort of context?

How 'bout this new paper in PNAS that used paleoprotoemics to prove that Neanderthals made jewelry too? 

Now, I'm not going to go putting Neanderthals down or anything, especially after this nice animation that greets me when I go and check on the new 23andMe features they've added....

...sigh...  You know, if  I was a more sensitive guy, I might actually consider that somewhat stereotypical caricature of what is apparently my ancestors a little insulting.  

Back on topic!  Okay, so I'm going to have to swallow what is left of my pride and admit that I don't get the central premise of the paper I'm talking about. This historical dig has been highly controversial. People have been digging stuff here for like 40 years or something and no one buys the radiocarbon dating cause they think the layers of deposition might be mixed up. So...what I'm missing here is how proteomics can make this clearer. But it made it into PNAS and got a nice review in Science so I'm gonna guess the experts on that side of things are cool with it!

What I do get is this -- the darned Neanderthals produced collagen that is different than ours -- well... yours, I guess!  Human collagen has tons of aspartate (D) in it. But neanderthals primarily substitute D for N (asparagine) --or vice versa, I read this on a boat today and didn't take great notes. And..you know...the caveman thing....

Okay! So that's easy, right? We can definitely tell if a D is N, that is a 20mmu difference or whatever.

So I'm super excited about this paper and then I get the PNAS paper and its and I keep seeing the words "MALDI-TOF" as I go through it. Which makes me just a little stressed because I'm afraid a paper getting mainstream attention might be looking for a 20mmu peptide difference with an instrument that is only accurate to 100mmu (if it is calibrated every half hour or so) but on page 3 of the supplemental info you'll find that they used LC-MS on a Q Exactive and I chill out. 

Like most PNAS papers, this one has about 100 pages of supplemental info, including some nice MS/MS spectra cut from PEAKS that support their D to N substitutions. This is a really solid and interesting paper all the way around. 

"Sweet earrings cousin Pat! You know who else made jewelry...?"

Cool summary of p53 structure/function!

p53 is a super important protein in cancer. Close to half of the tumors identified in the U.S. each year have at least one p53 mutation. p53 is also a pain for proteomics because, under normal (and some of the mutant conditions) its cleaved up by another regulatory protein so rapidly that we'll often not see even a single peptide from the protein in a global analysis.

This summary in Science talks about some of the hurdles the drug development people (trying to restore messed up p53) have to go through and is a solid summary of the protein, what it is and how it works.  As a free bonus you totally find out 1) that elephants don't get cancer and 2) why!

Want to do capillary/microflow proteomics? -- here's a great reference!

I hate nanoLC. I've hated it from the first time I heard of it, and I still hate it today. But its one of those necessary evils, like the IRS and alarm clocks -- and it currently still appears to be necessary for  doing proteomics.

The usage of nanoLC boosted the sensitivity of our LC-MS systems right through the ceiling. Some of the math out there suggests something like hundreds to thousands of times more signal versus analytical flow LC -- and that makes sense. But the mass specs are now hundreds -- maybe thousands of times - more sensitive than the stuff I learned to use, right?

Posts like this have appeared on this blog before, where I've worked with someone and we'd use analytical flow and get X hundreds or X thousands of peptides/proteins. What we haven't spent a lot of time on is microflow.

This great new poster that Alexander Boychenko et al., showed at HUPO 2016 shows what you can do with microflow LC (they call it capillary) versus nanoflow!

All LTQ and Orbitrap systems can use microflow -- here we're talking about 10s of uL per minute. Is it as sensitive as nano? Nope. But what kind of depth do you need in your samples? If you need comprehensive proteomics coverage -- you probably already have a nanoflow system. You could also do upstream fractionation to get the greater depth.

Do you need a lot more sample for microLC? You sure do! But...are you that sample limited? A plate of adherent cells in culture is going to give you something between 1 and 5 mg of protein depending on the cells and how you harvest them.

I'm not saying we can ditch the nanoLCs yet. Maybe we never will -- but if I was in a lab where I didn't have one this is a great app note for allowing you to do some light proteomics in between your normal small molecule workflows!