Saturday, November 24, 2018

How to optimize collision energy for massive numbers of crosslinks!!


Okay -- it might be time to put a page over there ---> for just crosslinking stuff. There is too much and it needs organized somehow. Time for another trip to the Room of Spirit and Time....

By the way, all this stuff isn't getting into the realm of redundancy at all. Crosslinking mass spectrometry does work. When it does, those spectra can immediately jump your research forward 10 steps. I've seen it happen around me at work multiple times this year thanks to data that I've personally generated for other people. However -- it's still got some serious limitations. The biggest one in my mind  is that the stuff that my group is actually doing....well....our stuff doesn't work at all. You still need to keep the samples very very simple. When we move the complexity up to even just one organelle -- blech....hundreds of thousands of MS/MS events? A dozen crosslinks the IONTRAP tells you are real. What's a 1% FDR of 100,000? Is 1% FDR of a crosslink MS3 even possible to estimate? Am I typing "1% FDR" repeatedly because I know that this totally incorrect, though somewhat conceptionally useful shortcut of confidence estimation that we say outloud all the time makes (some?) statisticians dream about murdering every proteomics person in the entire world so we'll be forced to start over with ones that they can teach basic statistics to?   🙉

Fun thought. What would a statistician do to get rid of all of us...?

What? Crosslinking? Oh yeah!

Here is the thing, though. When you see it work it looks like "man, just some minor tweaks will make this work. I swear." You feel so close and that is what makes it both so alluring and so frustrating.

Will this be the study to tip the scales???  I hope so!! 


I'm primarily using the MS3 based crosslinking stuff on the Fusion. We're now investigating a new method the great Dr. Bernard Delanghe taught our group recently that utilizes both MS2 DSSO triggered MS3 and ETHcD and it's showing some serious promise -- but we're all familiar by now with the pseudo continental divide --


You don't need to read much of Dr. Mechtler's stuff (or hear much of his accent 😉) to realize he's firmly European. This group is consistently showing us that if we use mass accuracy (and our heads) we can circumvent the need for multi-level fragmentation and slow, low efficiency fragmentation methods. And this study is another chapter in this growing tome. I'd title this chapter.....

MS3 DSSO crosslinking < HCD MS2 based DSSO Xlinking 

(with proper CE optimization!!!!)


With this paper in hand I now have a reason to head to lab on a gross Saturday morning and see if I can get some of these samples to give up the crosslinks WE KNOW are in there somewhere.....

No comments:

Post a Comment