Friday, March 17, 2017

Should you be using lower collision energy for PRM experiments? I was running my mouth again about how PRMs on a Q Exactive could beat SRMs for a QQQ and had to blow a weekend in lab running stuff to prove it to a bunch of skeptical people.

Caveats here for why I made this very costly dare (I probably only have a few thousand weekends left in my whole life after all...)

This researcher has only one peptide that he can use to confirm a positive outcome for this assay. One peptide. (Plus controls and whatever, of course).

There will be pressure for the LC-MS assay to be as short as possible.

The matrix is...whole...digested...human...plasma (or serum or whatever. A friend told me there was a difference yesterday and I still don't know what it is)

If you've got a protein you can get 3 peptides from for this, okay -- a QQQ might be the better choice for this assay -- but if you've just got one? I'm going PRM all day and never consider the QQQ.

I can't show the actual data cause I signed a thing that looked seriously scary. But I can tell you this -- there were so many peptides in the close mass range of this peptide in the digest on a 20 minute gradient that there was no way I could even trust SIM -- even at 70,000 resolution (max I had on the instrument I used) -- nope.  HAD to be PRM.

And -- when I was looking for fragment ions for my quantification (btw, I just extracted with Xcalibur and I believe it sums the fragment intensities rather than averages them -- but I'm not 100% - the peptides look great in Skyline as well) there was enough coisolation interference at with a 2.2 Da window that I couldn't use anything in the low mass range at all.

With this information I created the super-scientific scale that you see at the top of this post.  I really had to go to high mass fragment ions for specificity in my quan (and the best possible signal to noise!)  How complex is the matrix -- that with a 2.2Da isolation window there are smaller peptides you can't trust -- extracted at 5ppm...?

And, you know what? I could boost the intensity of these big fragment ions by dialing the collision energy back some.  Not a huge boost, but dropping the nCE down to 25 might have picked me up 10-20% in this particular assay for this particular peptide. (Your results may differ)

Let's check some experts!

I went to ProteomeXchange and searched "PRM" and downloaded some RAW data at random from a couple studies out there....and...I totally "discovered" settings I should have been using the whole should probably use a little less collision energy for your PRMs!

The first 2 studies I pulled...used...25! (PXD003781 and PXD001731). 2 other studies -- RAW files completed just as I was wrapping this up appear to have used 27.  We're at 50/50, but my peptide really liked lower energy.

Side note -- these samples were given to a lab that ran them on a QQQ that would cost this researcher MORE than the Q Exactive I used, LOL!

BTW, the  QQQ lost again. In ultra complex matrices where QQQ is going to lose the S/N game -- and you don't really need the 500 scans/second -- what you need is certainty that what you are quantifying is the correct compound -- my money is on PRM. And -- holy cow -- if you can save money getting a Q Exactive over a QQQ for the assay....

1 comment:

  1. This is a great tip, Ben, thank you. Also plasma is the liquid, cell free part of blood that has been treated with anticoagulants. Serum is the liquid part of blood after coagulation so doesn't have clotting factors in it such as fibrinogen. Kinda the same but different.