Wednesday, September 26, 2018

Splitting up TMT kits? Skip every other channel and increase your coverage!


We don't always have 11 samples to run. In cost per reaction though,  (at least with my old sales rep -- the new one doesn't seem to appreciate the level of discount I expect 💔😇😇) TMT-11plex is about the cheapest way to go.

If you don't have 11 channels, for example, you have 6 -- you can get a boost in your number of peptide and protein IDs by skipping every other channel.

If I have 6 samples I will pick an N or C variant for each unit mass and stick to it. For example

126
127 N
128 N
129 N
130 N
131 N

And make sure to not mix in any of the C variants from the same unit mass. To fully resolve 129N/C you need 43,225 resolution @ m/z of 200. This means you need to use 50,000 resolution on a Tribrid, QE HF or QE HF-X. Or 70,000 resolution on a QE or QE Plus.


Sooooooo slooooow..... Great data...but...sloooooowww....

If you skip the N/C variants for each unit then your reagent is essentially the TMT6-kit (with shuffling of the N/C)

At 1Da apart, it really doesn't matter that MS/MS resolution you use. I drop it all the way down (30 Hz on the Tribrid!). You aren't going to get 6 times the number of scans -- but unless you are fill-time limited (having trouble hitting that AGC target) you're going to get a lot more MS/MS, peptides, and protein IDs than if you'd mixed your N/C labels!

Worth noting -- if you are using Proteome Discoverer you probably want to create a new quan method for the channels you use -- if you want PD to normalize the data


-- if you aren't normalizing and/or imputing quan you probably don't need to worry about it. You can just hide those channels in the output report. What you don't want is background noise in your quan region (or a reporter M+1 isotope) being mistaken for your quan, it being scaled up as if you loaded only 1% of peptide in channel 128C and some noise being amplified 100x. Probably harmless in the end, but it totally looks weird and worries me that it might actually affect something.


1 comment:

  1. Just a minor correction: You're a bit off on your resolution calculations, it goes down with square root of m/z, not up. And 30,000 is enough for ratios up to 16: https://www.ncbi.nlm.nih.gov/pubmed/30220210

    Cheers, awesome blog - I like it a lot!

    ReplyDelete