SAPH-ire: Structural analysis of PTM hotspots!

Continuing theme alert!  What do we do with all of this proteomics PTM data?  Maybe we run it through this awesome new program from Henry Dewhurst et al., that they call SAPH-ire, or Structural Analysis of Ptm Hotspots.

What does it do?  Well, it is a way of looking at the post translational modifications you've uncovered in a a quantitative way and groups them to give you an idea of what they may be telling you.  Do I fully get it? Honestly, not yet.  But it seems really really smart and I think I'm on the edge of getting it.  Hopefully another coffee and it'll all fall into place.

I think this kind of thinking is critical, though.  Say I do a Byonic search and I end up with my proteins group of interest and in it there were 10 phosphorylations 16 lysine acetylations and 30 -odd other PTMs.  How do I get anything out of that?  If there was a way of clustering to say that "under these conditions you found more of these modifications" thats a big step forward, right?  And what if you could tell "under my conditions (drug treatment or whatever) that it wasn't a change in a certain individual PTM, instead it was that an area of the protein got PTMylated in some way..."  Maybe we're focusing too much on the individual modification rather than the fact that the region is active.

Okay. I'm another coffee in and maybe its just the caffeine euphoria thing but I think I seriously love this paper even though I'm going to need to keep thinking on it.  Check out this figure I stole from the Supplemental:

These are PTMs searched for in their proteins of interest (G proteins!) plotted against one another.  These are the protein "hotspots" that they focus on.  There are regions of these proteins that are active, but not all in the same way, or in a simple way.  We've got multiple PTMs occurring in combination in these areas and I've never seen anything that could provide this information before.

P.S., they validate the heck out of this thing.  I'm going to revisit this because I'm starting to lean toward thinking its positively brilliant.  Now, I just need to get my hands on it...

PhoSigNet: Can we make sense of all that phospho data?

So...you set up this elegant phosphoproteomics project.  You dual step enriched, you used a brilliant LC-MS method with neutral loss triggered ETD and now you have this fantastic list of 163 phosphorylation sites that are up-regulated by your drug.  What the heck do you do with it now?

Well...you can convert it to gene identifiers if you are in mouse or human and give it to Ingenuity Pathways Analysis (IPA) and maybe that'll get you something.  But that is gene-level data and its probably not going to help.  You can dig through PhosphoSite manually looking at the 163 sites that you found (or the 91 of them that have been annotated).

PhoSigNet is the ambitious endeavor of Menghuan Zhang et. al., to help us with this process.  They have taken the data from PhosphoSite and CanProVar and other databases and ended up with 200k phosphorylation sites (almost 12k of which have been validated in one way or another!)

Through an algorithm that they call ExpCluster you can upload lists of your quantified phosphorylation data and it'll try to figure it out.  Unfortunately, I'd have to dig pretty deep into my old hard drives to find a list of data to feed it (and I need to go to work!)

You can check out this awesome looking resource here.

New proteomics discussion board?

I think the list of things I want to blog about right now is expanding at nearly the same rate as the amyloid plaques in my brain!

This is a quick one I can check off the PostIt note(s!). A postdoc in our field (anonymous?) has proposed a new central proteomics forum.  I know we have a few here and there (ABRF, sharedproteomics, BRIMS) but they are kind of disjointed. This one is hosted by StackExchange. It has a friendly interface and can be directly linked to your Google or LinkedIn or other similar accounts.

StackExchange is big enough as a discussion board in general that it can be supported technically.  There is nothing on the site at the present but it might be another good way to get connected to people with the expertise to help you when you're stuck.

You can check it out here!

PIA -- Open source protein inference

Until top-down proteomics really reaches its potential most of us are going to be doing shotgun proteomics.  And it has this nasty drawback.  We can make a peptide-spectral match (PSM) but sometimes that PSM can be linked to multiple proteins.  Figuring out which one it came from can be hard, if not impossible.

We've seen some good new tools lately such as FidoCT.  PIA is a new one and its open to everybody and a bunch of search engines.  You can read about it in this new paper in JPR here.

Some of the highlights are that its got an easy interface and it can take data from just about anything.  I'm toying around with the web interface now and its got data from MS-GF+ and X!Tandem (and Mascot? I forget now, but I think I saw it, but I don't want to hit the "back" button on it) and some other programs already up there for me to mess with.

The number of metrics it uses to support its inference is a little overwhelming.  This is obviously a powerful piece of software.  The web interface seems to be more of a demonstration of the power it has and the true software is downloadable and fully scalable from single PC usage to use on servers and clusters.

You can check out the PIA website here.

Imaging MS of the scariest thing I've ever heard of

I am a huge fan of imaging mass spectrometry.  I've never really done it, with the exception of a training course in Bremen but I'll take any opportunity to visit people who are doing it and I'm a big fanboy of the research, so I know some about it.

In a fantastic use of this technology, Brian Flatley et. al., took a look at a cancer biomarker called S100A4 and its distribution through tissue of something terrifying that I had no idea existed until I downloaded this paper.  (P.S. There were no tissues of this type in the huge cancer genomics project I did years of QC analysis on).

Working with histologists who could identify the cell types by microscopic evaluation they could break these tissues into distinct areas and then looked at outliers that changed quantitatively in different slide areas.

It really is just a stunning paper visually with some great work clearly lined up. I hope with work at this level that this disease (and all cancers in general, of course!)  will soon be something that none of us ever hear about again.

Just when you thought glycoproteomics couldn't get any more complicated....

..this study happens.

Well, maybe, I'm exaggerating. I think we're all getting a good feel for just how crazy complicated and essential our understanding of glycosylation is to biological processes.  But this is a great study to underline this fact as well as demonstrated (to me, at least) a different way of visualizing glycosylation

In the study they took mouse livers, enrich for glycosylations at the protein level,  digested everything and did dual stage HCD + product ion triggered ETD on an Orbitrap Velos.  The resulting data was searched with Protein Prospector using "an iterative searching strategy" described in this paper I haven't read (yet!).  Essentially it looks like they are doing a deltaM style search with a 2kDa modification window.  I'm assuming the iterative part is a data reduction step.  However, I'm thinking if I sucked it up and gave up the bandwidth I could pull this off with Byonic or ProsightPC.

The output is this crazy awesome histogram with glycosylation masses versus frequency.  What a nice way to summarize this data!

The paper goes further and they use some advanced versions of gene ontology stuff to figure out that the patterns of glycosylation are organelle and tissue specific. 

High resolution discovery ported to a clinical assay for prostate cancer recurrence!

This study by Claire Tonry et. al., is an awesome example of porting a biomarker discovery project directly into a clinical assay.

In this study they followed patientswho had treatments which pushed prostate cancer into remission.  They followed the course after this and identified biomarkers from the unfortunate guys whose cancer recurred.  Their comparison gave them 65 sweet new protein biomarkers that they were able to seemlessly convert into a beautiful new clinical assay for MRM.  At 65 biomarkers, they could have used the discovery Q Exactive as the validation instrument (we're not seeing this enough yet!) but I'm certainly not going to complain.

This is a great study that shows how we can use these technologies complementarily (apparently..not a word...) to benefit patients.

MCP Rules again!!!!

How excited am I about this?  So excited that I'd make a bad illustration of the MCP logo and celebratory fireworks?  Yup!

Wait. What are you excited about again?  Only about this statement in MCP!!!  It is called "On Credibility, Clarity and Compliance" (I really wanted to throw in another funny C word but I couldn't come up with one.  Google suggested "cacodemomania" which is the fear that you are possessed but this is far too cool and serious.

I've always had a soft spot in my head for MCP. And not just because now that I'm on the dark side I still get access to most articles they publish.  I have appreciated MCP because of the focus on quality.  I was stunned, annoyed and possibly a little angry when MCP had to drop some of their requirements for publication, particularly the requirement that all RAW data was publicly accessible.  Seriously, though, what do we do with all this data?  We filled the venerable and powerful Tranche servers until they hardly worked. I keep just dropping more hard drives into my home PC to keep up with all the data that just the people I work with in my day job generate.

But you know what?  MCP has the straight up intestinal fortitude to say "you know what? Yes, the data is getting bigger and its going to be increasingly hard to make public, but in order to make sure the stuff in this journal is AWESOME we are going to make it required to make that data available." Is it harder for the authors?  Yup.  Is it harder for everybody at MCP?  Of course.  Is it the right decision? Hell yes it is.

Serious serious kudos to the editors for this.

ExpTimer -- automated alarms for all your benchwork

Yes, I have a list of 90 things to write about from ASMS, but this is really cool and I've got 2 minutes.

ExpTimer allows you to type your whole experimental protocol into your timer and get an automatic alarm when its time to (for example) add your iodoacetamide or something.

Nifty, right?!?  You can download it here.   And read about it here.

Shoutout to @PastelBio for the legwork!

Matrix assisted ionization for Orbitraps

If you are at ASMS and are want to know an awesome booth you should look in on?  I suggest the one with the great big "No MALDI" sign.

What they have is matrix assisted ionization that doesn't use a laser and connects right up to any vendor's mass spec.

If you aren't at ASMS you should check this out.  Here is their flier (sorry for the potato quality):