Low number of IDs

[Michael Porter]

I gave MaxQuant.Live a go but so far I'm disappointed. I tried the boxcar method and my resulting mgf file was 85% smaller than if I had just run a top 20 DDA method. I am running Tune 2.9 on a Q Exactive HF and I used the default boxcar settings. My LC is run separately from the MS, so the only difference is that the dummy MS method is set to run for the entire duration of the run rather than just a couple minutes. However, I don't think that caused any problems because MaxQuant.Live still took over the entire acquisition. I do understand that the number of MS2 spectra may drop because the boxcar scans take a little extra time, but I'm just not seeing how an 85% drop in my MS2 spectra lines up with the original claims of getting 10000 proteins from a single 100 minute run. Is there some setting somewhere that I am missing?

[Ryan Hill]

Hi Michael, 

  I just processed some data comparing the two methods and came out with the same result. It makes sense based on the duty cycles being comparable but so much more time being spent on the two boxcar scans in addition to the suggested full scan. One discrepancy from the paper compared to the default settings I noticed was the dramatically lower maxIT (250 default, 20 in the paper). I don't have that instrument to try this for another week or so, but will let you know if some playing helps at all. The initial results look promising in terms of number or MS1 peaks identified, so hopefully we can keep this sensitivity increase while garnering more MS2 spectra. Let me know if you came up with any solutions in the meantime.

Thanks!

Ryan

[Vladimir Gorshkov]

Hi Michael,

To get those number of identification peptide library was used in BoxCar paper.

p444  Deep proteome quantification with BoxCar

"Given the greatly improved dynamic range and number of detected features of the BoxCar method, we asked whether comprehensive proteome characterization could be reached by combining single BoxCar runs with a peptide library24,25. In such an approach, peptide identifications would be transferred from the library using the ‘match between runs’ feature of MaxQuant26, whereas the quantitative information was provided by BoxCar single runs (Fig. 4a and Supplementary Note 1). "

So, the quantitation is performed by BoxCar scans, while the identification are produced elsewhere.

Best regards,

Vladimir

[Witold Szymanski]

Ok, I am confused,

How can you have a quantification performed by BoxCar when you have way less MS2 features in comparison to a standard DDA method? We also get way less MS2 events in our tests and less identifications in total, also when using MBR with other normal DDA runs. 

Greets

Witek

[Ryan Hill]

Vladimir, I had missed in the original paper that all deep proteomics identifications were purely from MS1 peak matching and RT matching. Thank you for clarifying!

 Witold, the quant is calculated from the precursor abundance and not the MS2, and you do get far better MS1 ion accumulation in this method. 

It seems limited to being useful in deeper proteomics to when you work with a single or just a couple of sample types that you can have a deep spectral library available for. 

Thanks for the discussion all!

Best,

Ryan