Thursday, September 27, 2018

Positional phospho-isomers are a problem -- get a Thesaurus!!

(Image stolen from [please don't sue me])!

I've been dying to talk about this one since seeing a talk sometime in the spring about it!

Did you know that phosphorylations are commonly associated with phosphorylations on amino acids right beside them or just a few amino acids away?!?  I didn't, but I've asked a bunch of biologists and they said it's true.

This is one of the many cool insights that you could find in this new preprint on THESAURUS!

Before you get too excited -- Thesaurus is for DIA and PRM data. Wait -- You're more excited?!?!

(Groans.....okay...last one, probably....)

Thesaurus is software. You might have guessed that from a couple of the names on the paper. And it -- okay -- figure 1 is awesome and explains it better than I possibly could.

Somebody is good at making flowcharts. The end result of running through that logical circle is going to be a test of whether phosphorylation at E or F is the best match ---

Okay ---- last stolen picture for this post -- but this is the ABSOLUTE COOLEST PART --- what if it is both of them? Because it biologically makes sense that it could be. No -- not phosphoRS doesn't have enough information to discern which one it is and gives you 50/50 so you just report both --- like, biologically it can and does totally happen that you'll get an almost perfectly co-eluting pair of peptides that is both phosphopeptides E and F (obviously not in the example above, but it really does happen a lot (proof in this great paper!)).

But this is the ABSOLUTE COOLEST PART (wait. I said that. I'm excited.) in modern dd-MS2 -- we skip the second one! Almost always!  We're so certain of our massively improved peak shapes and the efficiency of our instruments in making an ID on the first fragmentation that most of us use dynamic exclusion to trigger at some (by historical standards) ludicrously low peptide intensity -- and then we exclude peptides of that exact mass from being fragmented for huge amounts of time. So if there are 2 positional isomers eluting at almost the same time -- we don't see it.

Is it possible that our improved methods and instruments is actually decreasing our phosphopeptide ID recovery? Yeah, it totally is.

EDIT: Forgot this part --> In DIA and PRM you are constantly acquiring MS/MS spectra for your mass range in a cycle. So you can see fragmentation patterns of two almost completely co-localizing phosphopeptides and Thesaurus can help you identify them!

I think DIA has kinda been floating around looking for something that it's good at -- or better at than dd-MS2 -- this might actually be that thing.

Thesaurus is shown working in conjunction with Skyline throughout the paper. It can also function as a complete stand-alone and you can get it here.

(They got this wrong, btw....🙉🙉🙉🙉🙉🙉🙉!!!!!)

No comments:

Post a Comment