Sunday, July 22, 2018

CPTAC 3 -- First paper already! How the TMT protocol is set up!!

CPTAC3 (might not be the official designation, but that's what I call it) kicked off last year. What I knew about it -- full standardization. All labs have the same LC, same mass spec, use the same columns and chromatography, and they moved up from iTRAQ 4-plex to TMT 10 or 11.

Now we can all know the entire workflow!!

Interesting points? Despite having Orbitrap Lumoses, CPTAC is sticking to MS2 based TMT quan with a super interesting side note ---  they recommend running with Advanced Precursor Determination (APD) off. And suggest that the Q Exactive HF-X shouldn't be used for TMT quan because the APD can not be turned off...?

Interesting, right? But it kinda makes sense, if you are running at a lower resolution at MS1 (60,000 for CPTAC3) maybe you don't resolve as much of the lower abundance peptides that have higher coisolation interference. If APD is helping you to find more lower abundance peptides in the noisier ranges maybe those aren't going to be peptides with good ratios anyway.  I dig the logic. I also dig the 0.7 Da MS/MS isolation window.

The study also suggests a really high degree of variation in HCD collision energies in their lab's Lumos devices. The actual settings table uses HCD CE of 34, but they emphasize CE optimization.

What kind of results are they looking at? About 10,000 proteins from xenografts with around 7,700 of them coming from human and the rest from mouse cell/protein invasion. They also show some really sharp correlation numbers suggesting their MS2 TMT quan is working just fine for them!

All the RAW and processed data is, of course, available at the CPTAC portal here.

Saturday, July 21, 2018

EuPA Practical Proteomics Day 5 (late) Recap! Kinase cascades and Crosslinking!!

EuPA Practical proteomics ended with a bang. 

Boris Macek talked about how to study PTMs -- and how to quantify and normalize them -- and then showed an awesome study that is in press right now. I'll be noisy when my Google Scholar alert dings me that it's out.

Then we went into crosslinking for the rest of the day with Juri Rappsilber (with Lutz Fischer for the hands on section.

What's that above? That's just some crazy amazing crosslinking report I generated in the hands-on workshop on day 5 from some BiorXIV data we pulled down using ---



and the released at the workshop but not quite published yet(?) xiVIEW(!!) All of these tools were developed by the Rappsilber lab and --together -- they're ridiculously powerful.

P.S. This is also the team that developed the xiNET online interface -- which I've been using for downstream interpretation of every crosslinking experiment I've done this year. (If you buy the $500 Xlink nodes for PD 2.2 or 2.3 it will automatically output the data into the xiNET format.

This brings up a very important point about all of these amazing tools this great lab is giving the world for free -- they've also enabled an amazing degree of compatibility. It looks to me like I don't need to use xiSearch at all if I don't want to. It looks like I can take my crosslinked data from any tool that I use and then put it into xiFDR (once i get the format right) and clean up my data with their extremely well-thought-out FDR calculations. Then I can take that data to the next step.

I sat through this workshop in something akin to slack-jawed amazement. And when I did finally say something it was a loud profanity (sorry....). I'm doing a decent amount of crosslinking right now -- pretty much successfully -- but I'm handing off PPTs and Excel sheets to the people who gave me the samples -- these tools get you all the way to the biology and is something we'll be using for sure!

Friday, July 20, 2018

EuPA Practical Proteomics DAY 4 (LATE!!) INFORMATICS DAY!!!

For the first time in my career -- I was the Session Chair!! For INFORMATICS!! PERFECT!

Oh. Umm.. Ugh...I REALLY WANT TO TELL YOU ABOUT Sebastian Dorl's talk. I'm not sure I can. It isn't public and it isn't published. Okay -- soon.

Lecture 2: Robert Van Ling -- 2 meter long nanoLC.columns that run 100 bar backpressure. Actually -- let me write this again. 200 centimeter columns -- 100 bar backpressure.

Not joking. A big thing here was that C-18 UHPLC beads in long columns, while functional, is kind of 1990s technology. The real separation scientists have moved on.  PharmaFluidics uses pillars on chips rather than beads in capillary tubes and can achieve separations that make what I'm doing seem very...

...they're not the cheapest things in the world, but maybe if you see the data for what monolithic columns can do in nanoflow (he showed 1,000 injections with no peak shifts) you might not mind.

Lecture 3: Boris Macek

I might have went all fanboy on this one. Dr. Macek IS the first author on LOCKMASS injection and the first paper I know of using topdown MS3 -- and is responsible for most of what I studied in school regarding phospho signaling in B. subtilis (yeah -- we knew other stuff before those papers, but the model organism for gram+ bacteria -- he was in Mann's lab when they characterized that stuff.)   His lab has done a ton of work recently on elucidating other model organisms that we thought we knew -- and there is something huge in press right now (Science Signaling, I think). Boris gave a VERY clear and very good lecture on studying PTMs with mass spectrometry. There was some high level cutting edge stuff here -- and not every student here had been doing this for years. GREAT TALK.


MSAmanda 2.0 and CharmeRT and Elutator and all the other free nodes that you can use in PD without paying anything. These are all feature prominently here. She also talked about the EuBiC winter school. I'll post details when I have them.

Worth noting -- I'm using CharmeRT A LOT. I think they underestimate what you get from second search. I'm commonly getting 2x PSMs when I turn it on. Yeah -- it could be faster. It's hard to search every MS1 peak in your isolation window -- especially when you've got Fusion #14 and you can't isolate less than 2.5 Th (don't ask me -- apparently it can't be fixed)

Lecture 5 and Workshop: Sebastian Virreia Winter -- Sebastian talked about EASI-TAG (the paper is out now) and did an awesome hands on workshop with MaxQuant and showed us some of what is coming in MaxQuant live -- amazing new instrument methods for the Q Exactive HF-X!   While he was here we made him show us some BoxCar methods and data and answer some tough questions from the audience on MaxQuant data processing.

EuPA Practical Protoemics DAY 3 (late) recap!!

Okay -- Day 3 is going here but is kinda sparse. I did something stupid (surprise!!! Ben at almost 40 can't hurdle everything he wants to anymore -- {"runner's high" makes me a little crazy in the morning -- seemed like a great idea at the time} and I missed a lot of talks while I focused on remembering how to breathe.  They were from some Apps scientists from Thermo (Sega Ndiaye and Peter Mowlds) they seemed really sharp and like nice people.

Then I took the stage after all the NSIAIDs kicked in and talked about QC/QA and online diagnostics our facility is working on. Slides will be available soon.

Yeah -- I know -- but we vetoed my talk on downstream bioinformatics because Europeans want to talk about QC. I talked about what we're setting up in Frederick which is based on work at the instrument side from Tara Schroeder and Lani Cardasis and data monitoring, which is 100% based on Michael Bereman's work. People in Europe didn't boo me when I wanted to talk about QA and QC!! That's a big deal. I stayed for Ilaria Piazza's talk (covered earlier in the week) and slept 14 hours. Did wonders for my spine and all the weird noises it kept making.

EuPA Practical Proteomics Day 2 (late) Recap!!

EuPA day 2!! PTM day!!! I'll update this post later when the lectures (that will be available) become available!!

Lecture 1: Kris Gevaert (lab link here)

TERMINOME STUDIES!!! Okay -- I know what terminomics is -- and I'm excited that other people do it and I don't have to yet. If you had to do terminomics -- maybe this is enough to give you nightmares. What if you had to do it in plants? What if you had to do proteogenomics and terminomics in plants? Honestly, this might be enough for me to leave the field.

They just published something like this. Everything else has to be easier. Holy cow...mad mad props to this team.

Lecture 2: Glycoproteomics!!!!

More brand new techniques and tools from Dr. Huber. What about advanced statistics in glycan characterization and localization? This paper JUST came out detailing part of this workflow!

What's better than a great new tool in the fastest growing branch of proteomics? Having a great name for it! This tool is called MoFi. And -- just to make sure with 100% certainty that you never forget it. Just once -- picture this legendary actor saying it.

Lecture 3: "Stories on protein biogenesis, acetylation, processing and stabilities". Sound like a great title?  Yeah!  Fingers crossed those slides get made publicly available, cause there is NO way I can do a topic this complex justice

Final talk? was me. Feeling 100% outclassed and not anywhere near accomplished to be sharing a stage with these people.

Lecture 4: BoxCar --> how the method works and why it is important. Give me a couple days, I'm writing something up and I'll have slides available showing how you can probably be doing BoxCar with no special software (if you've got a QE, it looks like we've got it running now, but give us some time to verify. We lost some pump seals on the QEs while I'm at the best meeting ever!!

Today cut a little short because we had the student poster session. I lucked out and got to be head judge. No joke, ya'll, the next generation of young scientists are SCARY. I was like "well...this is smarter than anything I've ever come up with, and -- ummm -- so is this...fuuu..." Almost all of it is work in progress, (set my Google Scholar alerts, for real) but I can point you toward one AWESOME tool that is live now (a lot will come later).

Are you using MSFragger? Wondering what to do with all the data that it kicks out?  What if you could download an awesome Python tool that would statistically rank that noisy output and help you make sense of it?  I mean this in the kindest way possible -- but MSFragger output is kind of imposing -- backing the output up with statistical grouping seems like a GREAT next step!

You fancy smart Python people can check out Julia Bubis's AAStats tool on her bitbucket site here. She's part of the Gorshkov group that recently gave us IdentiPy.

After the poster session we took a tram ride through about 4 million awesome historic buildings in Vienna and then had dinner on the Danube!!!  (Sneaky picture I took of the students and instructors from a distance. It was way nicer than this).

Thursday, July 19, 2018

EuPA School on Practical Proteomics 2018 Day 1 Recap!!

There is no way I'm going to do this amazing meeting justice, but I'm going to try. You might not believe me, but I learned more at this small meeting in Vienna than I did at ASMS 2018. Karl Mechtler and his team invited an amazing group of speakers doing the coolest stuff that anyone is doing in proteomics right now and -- after being amazed by their work -- I could just bug them during lunch and at dinner (and maybe after).  I'll be writing lots about the stuff I saw here as this stuff is gets accepted.

Day 1!! TMT day!! (Late recap!)

How does the highest throughput proteomics lab in the world do what they do?

Who am I talking about? The Haas lab at MGH. I've been in a lot of mass spec labs. I work in a really big one that runs more samples per month than I'd ever believe possible -- Haas lab is tearing through more proteomes than anything I've ever seen. We got a morning of logic into using TMT SPS MS3 to get near complete proteome coverage on 10 human cell lines every 24 hours or so.  In the afternoon we tore through the practical day-to-day operations and challenges of trying to complete thousands of proteomes per year. This lab is linked to a lot of papers -- and with this kind of throughput -- we've only seen the tip of the iceberg.

Christian Huber was the one break from this intensive TMT crash course. Dr. Huber delivered the best separations chemistry course I've ever seen. I suspect he might teach this at his University. You can check out his site here.  If you know anyone in the small molecule world you're probably aware of how primitive mass spec separation technology is for proteomics.

It was still a wake up call when Dr. Huber called out some of the more seasoned people in the room and asked them if they remembered PepMap....which...I...umm....totally still use.....and some people I respect A LOT are in the room chuckling because they consider it an artifact of history....this leaves me wondering....

 ....use denial when you makes you feel better than thinking you're wasting loads of peptides with lousy chromatography.....

Day 1 needed to end a little short since I'm presenting data on day 2 that is currently being processed....fingers crossed!

Wednesday, July 18, 2018

Chemical Proteomics --- I promise you want to check this out!!

I somehow lucked out into getting an invite to the AMAZING EuPA School of Practical Proteomics in Vienna. A lot more will follow as I wrap up what is one of the best meetings I've ever been to.

I've somehow missed this study (this year has been busy...)  I just saw the author speak on it and I promise you want to check it out. I've already sent it to some of my collaborators.

What if we could measure protein-protein communications and protein-metabolite interactions on a global scale -- RIGHT NOW?  No way, right? The technology isn't there yet!

This methodology is sneaky simple elegant and ridiculously powerful...hey...

....who makes these things....
....some picky jerks in some obscure journal called "Call" or something liked it too!

Monday, July 16, 2018

HTRA1 -- The most important protein you never heard of?

Have you ever heard of this HTRA1?!?  If not -- this paper in press at JPR suggests that you (and I) should review the (scant) background literature on it -- or just read this great new study!

What is it? be certain I'm not sure we knew until now. Seems like the literature goes back a few years showing it's important -- but this study shows it's critical -- and what it's doing in our cells!

What's it linked to?
Little things like
-Macular degeneration

You know, nothing serious...but try Google Scholar digging through and wondering how you never heard of something this important. There is SO little. And this group decided to fix it with --- FLOW CELL SORTED QUANTITATIVE PROTEOMICS!! (Yeah!)

EDIT (7/19/18) -- After spending more time understanding the paper. Flow cytometry is being used to verify the cells are in the correct states in their cell cycle. I don't believe active sorting of populations is in play.

They synchronized 2 cancer cell lines, including a modified version of SW480 where they could finely control the amount of HTRA1 protein expression (didn't check details, but published elsewhere). This is where the model gets cooler. They block the cell cycle (with thymidine) using a really interesting 2 stage blocking technique I haven't seen before.

It looks like they block the cells at one stage (the synchronized cells can't continue about their usual business of dividing. They just stop all that stuff.) Then they wash it out in some cells and block the cells again at the next stage in the cell cycle. Sound hard? Sound like the matrix is getting pretty big? You're right on both counts!!

This great study doesn't stop there. They use another checkpoint inhibitor or two. What they get is a big picture of what different levels of HTRA protein does at different points in normal cell cycle stuff. Because the cells are syncrhonized and stopped all together, they have a big population of cells all doing the same thing. Messing with any protein in this system that is linked to cell cycle progression or maintenance is gonna have effects somewhere!

Wait. What? They also throw in some DNA intercalators? This study is big. Wow. Next time someone has any kind of a serious fundamental question about DNA damage or cell cycle checkpoints pull up this paper and --- boom!

(No joke. Go talk to someone who does Radiation Oncology. They'll be really impressed!)

Once this large matrix of cells at all these checkpoint stops were acquired, proteins were extracted and digested and ran on really really long nanoLC columns? I'll need to check on details. I just talked to someone today with a 200cm nLC column I might try, so maybe 380cm isn't actually that crazy. And all spectra were obtained on an Orbitrap Elite. All processing was MaxQuant LFQ and Perseus.

What do you get out of all this work? What about proof that this protein I never heard of appears to be linked to regulating as much as 1/10 of our proteome? And if you adjust this by a very crude estimate of the proteome coverage vs theoretical proteome -- what if it looks almost half as important as the cell cycle indicators that scream "STOP DIVIDING SOMETHING BAD IS HAPPENING TO OUR DNA". For real.

Elegant experimental design -- really clear output -- mysterious LC conditions? Great paper you should check out!

Sunday, July 15, 2018

QC-ART -- Real time(?) proteomics QC in R?

Credit for this link goes to @UCDProteomics. It's awesome to be mostly done with a QC proteomics talk and then something totally new and amazing pops up in your Twitter feed!

I can not guarantee I get this. I'm also not 100% sure I'm cool with something as wildly open source as R directly interfacing with my mass spectrometer -- but ---

This might really be worth checking out -- particularly if you've got an R BionformaGician around anyway and you haven't instituted a full out QC protocol on all your stuff!

They Shiny App interface looks really super friendly and non-intimidating.  The number of variable assayed may not be any more than what we're currently implementing (sProCoP/AutoQC), and I should probably do some stuff rather than digging through details I probably wouldn't understand anyway, but PNNL has a tendency to make some really good software -- maybe not the most dumb person (me) friendly software, but I bet this is worth taking a look at!

You can download QC-Art here!

Saturday, July 14, 2018

FAIMs is back?!? And putting it on an Orbitrap improves shotgun proteomics!?!?

For just a second I started this paper -- Pierre Thibault -- FAIMS and thought maybe it was 2005 and I had to check a mirror --

Nope -- definitely not 2005 --

You, too, might feel like you've stepped into a time chamber when you see some definite suggestions that FAIMS might finally be back -- and it should be if we can get results like these!

It's in open in press right now!

This paper is really big and deserves more time than I can spend on it this morning and should really be checked out.

Get this -- they do TMT-10 plex (this is FAIMS on a Tribrid). TMT-10 plex can't be done on another ion-mobility proteomics instrument that's getting a lot of attention (despite resolution claims it can't, resolve the reporter even our Orbi XL can) AND -- maybe even better -- it looks like they get better results than SPS-MS3 based TMT quan (less cycle time hit? with FAIMS? what!??! I know!)

Can I get FAIMS now? Doesn't appear so, but maybe soon??

Shoutout @KarlMechtler for tipping me off to this!

Friday, July 13, 2018

A really insightful analysis in paleoproteomics!

I'll be honest, I don't quite understand the paleo part of this study OR the proteomics part! However, my slowly caffeinating brain still realizes that what I do get suggests that this is really cool and has implications for what I do every day!

Besides the obvious WOW factor that this is a proteomics study in an Evolutionary Biology journal(!!) there is still a lot of insight here even past the paleontology part.  Dr. Welker is drawing conclusions from proteomic analysis of both distant relatives and very close ones (humans to chimpanzees) using different search techniques to show where they do and don't have power to make connections.

While most of us aren't doing paleoproteomics, almost all of us are searching proteomics data against protein FASTA files that don't contain the exact sequences of the organism you just lysed and chopped up with trypsin. Individual proteomic variation like single amino acid variants (SAAVs) are undoubtedly having an effect on just about every run we queue up. Maybe error tolerant searches like the ones used here are the temporary fix we need until proteogenomics gets easier (and sequencing gets a little cheaper)! 

Thursday, July 12, 2018

rawDIAG -- Rational shotgun proteomics optimization!

I'm not 100% sure that I appreciate the implications. Rational optimization?!? What is the alternative? Only some small percentage of us are randomly pushing buttons.  I'm joking about one or more things, possibly....

However -- I'm 100% that I love this idea and new tool!

Even us old seasoned mass spectrometrists can fall into the optimization pit. Building parameters and changing things more on feel and hunch and previous experience. It might totally work out -- especially if you have loads and loads of experience, but even then -- wouldn't you feel way smarter if there were concrete facts behind how you designed the experiment? In the absence of facts, what about fancy statistics?

Before you move on to looking at something smarter than this site because it says R package (gross), there is also a GUI for this! They even blast through some real world data (hundreds of core lab generated files) to show that this thing knows what it's doing!

Tuesday, July 10, 2018

Some 2018 MaxQuant Summer School Videos are up already!

And here I was thinking after a long but very great day that the dogs and I would watch some PandR -- and I get a YouTube alert that the MaxQuant summer school videos are already up!

If you right click on these videos below it will take you directly to YouTube where you can better control the screen size and resolution. Once you go to one of them, you can find the rest.

Here is this year's intro (and BoxCar)--

This one is...ummm...intimidating...? But really shows what you can do with Perseus if you know what you're doing -- Network analysis. No one ever said this software wasn't crazy powerful....





....but it can get a little complicated still....

This year's program can be downloaded here and that should help to find the lectures as they continue to be uploaded. I thought I'd link them all here, but then I thought...

Wednesday, July 4, 2018

ProCal (JPT Retention Time Standardization Kit) m/z

This might be something I'm leaving on the blog mostly for me so I don't have to look for it again, however, maybe someone else would find it useful?

If you are using the ProCal (now called the JPT Retention Time Standardization Kit) you know what I'm talking about and have probably already done this yourself, but if you lose your Excel spreadsheet with the m/z you can download mine here.

If you don't know what I'm talking about you should check out this paper!

Retention time, mass AND COLLISION ENERGY calibrators.

They are a little expensive compared to some of the other QC reagents (10pmol of the 40 will set you back $99) but the ability to easily verify that HCD CE 30 on the Fusion still looks the most like HCD NCE 27 on the HF...?..this is the easiest way I know to verify it!

Monday, July 2, 2018

XRNAX -- The ultimate methodology for studying RNA/Protein interactions?

I'm seriously beyond impressed with this new study at BiorXiV. What an amazing amount of work in a critically under-developed area!

I feel like this stuff comes up all the time. How do we study what is happening between the RNA and protein. If you're thinking -- bleh -- who cares -- didn't Rosalind Franklin and some guys who lucked out because some crazy alcoholic congressman hated Linus Pauling solve all the DNA--> RNA --> protein interaction stuff? We have the central dogma, right? We're good to go! 

Easy to think that but -- it turns out that it's way way way more complex than this. You know how they teach us there is like 4 nucleotides?

.... As of 2006 there were over 100 post-transcriptional modifications of RNA nucleotides known and I know 2 people that are working on studying new ones and whether they are biologically relevant. I only learned this recently and I was inconsolably depressed about it for like 4 minutes. For real. That's messed up.

Okay -- so with that in mind. Let's go into a less comfortable framework. Maybe we don't know ANYTHING about how protein and RNA interact, why they do it, and what happens when they don't do it right?  What the Heck do you do when someone asks you to study these things? If it's an email, you could probably pretend it went to your junk filter for -- what? -- 5 weeks? That's fair, I think. If you've got an office, though, you're probably going to get an appointment request by week 6.

What do you do? Schedule and appointment for 2pm and take a Xanax at lunch? (Obviously -- your legally obtained, doctor prescribed medication -- I assume having an office automatically qualifies you for some sort of anti-anxiety medication. Everyone knows where to find you now!)



This paper is huge and kind of intimidating. It's in BioRxiV now, but this should be in Cell or something equally impressive soon, I'm sure. But check this out. You can go to and the website walks you through something huge and scary like RNA/Protein interactions -- unbiased -- ridiculously powerful -- and it starts like this --

It's got photographs of everything. All the steps. All the materials and methods are so clearly described and pictured that I could probably do it.

I bet that's the clear stuff!!

It doesn't stop at the sample prep. The different levels of experimental design depending on what your question is. Do you want to know exactly where the nucleotides and proteins interact. Do you need something focused or quantitative global (SILAC)? its' all in here.

Applications of this XRNAX are described in the paper and at, but I suspect this is just the very tip of a huge iceberg. There is so much that we can do with this. There are a ton of fundamental biological questions to go after -- you can't tell me that when cells are irradiated or blasted with free radicals that it just busts up DNA -- other things have to be affected -- but DNA is the only thing we've ever really looked at (cause it's way easier). XRNAX gives us a whole lot of ways to look these complex interactions that I don't think we've had before.

While I'm just rambling. There are lots of other gems in this study. This is the first time I've seen the awesome MSFragger in action (indiscriminate PTM IDs -- they do +1000Da!!). I should probably stop typing about this, but -- wow -- you should really check this out!

Sunday, July 1, 2018

CharmeRT -- Time to get a lot more peptide IDs with advanced second searching!

Hey. Could you use -- I dunno... --60% more IDs in some of your datasets?!?  I'm going to assume you said "heck yes I could, but that would be crazy to ask for."

Say hello to CharmeRT!

I've been looking at this since it came out on Thursday -- and I still don't get all of it.

What I do get? Just about every time we fragment a "peptide" in a complex dataset, we actually fragment the thing we want to -- and everything within a dalton or two from it. You don't even have to look hard in your data to find really good peptides with tons of background behind them.

Random dataset I just pulled, sorted in terms of highest XCorr and highest coisolation interference.

Literally ever b and y ion was detected for this short peptide. I'll go way out on a limb here and say that  it's probably this peptide sequence. But look at all that other crap in the background!

Look at it's MS1

The target peptide and it's two isotopes are there -- but it looks to me like there is another peptide in there, right? We just want the one, but I fragmented everything in there that is yellow.

Could we go in and remove those nice matching MS/MS fragments above and re-search this spectra knowing that something in that yellow MS1 isolation window was the parent peptide?

And that's what CharmeRT is doing. Now -- this is where I get a little lost. It's more complex than this because it utilized retention time calculation through Elutator in some way to aid in the scoring. How cool is that? My guess is that on the second round you have to open up your MS1 tolerance -- so anything to help with your confidence has gotta help. Retention time is a great place to start!

The problem with retention time calculations is that you might be doing something really weird with your chromatography. If you are handpacking with noble gases while underwater or something and you just thought that this wasn't for you -- you just need this thing!

The Elutator RT trainer lets you load up some of your own data from peptides that have eluted off that mix of 80% C-18 and 20% Camel Dental Plaque that you know does the best job of getting those hydrophilic peptides -- and lets you create the training dataset that takes this into account for your lab and your conditions.

You can get all of this right now here!