Peak quality score for small molecules bwidner  2020-03-17

Hello Skyline Team!

I am working with small molecule data, and I would like to put a number on chromatographic quality for each peak. I would like this score to only be for MS1 peaks (or MS2, if I decide that's what I want to look at) and to include things like peak shape (i.e. gaussian), signal to noise, agreement with predicted retention time, etc. I have not so far found this in Skyline.

I looked over the Advanced Peak Picking Models tutorial since it includes scoring, but I do not think this is exactly what I am looking for. It seems more complicated than what I need; perhaps I am overlooking something?

I am new to MS and get a little lost in the terminology, especially when it is peptide terminology, as I have never worked on peptides. My apologies if I am overlooking something obvious, and thanks for your assistance!


Nick Shulman responded:  2020-03-17

You can export a .csv file containing all of the peak feature scores using the menu item:
File > Export > mProphet Features

Some of those feature scores are related to peak shape, retention time difference, etc. (I am not sure exactly what the "peak shape" score is based on-- I would need to read the original mProphet paper in order to say whether it has anything to do with the peak being shaped like a Gaussian, but maybe someone else knows).

Is this what you are looking for?
-- Nick
bwidner responded:  2020-03-18
Hi Nick,

Thanks for your response. I looked at the mProphet paper, and I dont think this is what I need. They define shape as, "The ‘shape score’ is a measure of global shape similarity based on the same cross-correlation matrix as for the coelution score. We used the property of a normalized cross-correlation that an apex intensity of 1 indicates perfect peak shape similarity and smaller values indicate noncorresponding peak shapes. The average of the apex intensity in the cross-correlation functions constitutes the shape score."

If I am reading that correctly, 'shape' just tells you whether the group of peaks all have the same shape; it does not evaluate the quality of the shape itself.

lasse responded:  2020-03-23

Don’t know if this should be a separate thread, but I think we have a similar objective. We are interested in the chromatographic performance of LCMS runs, and I am trying to decide if Skyline is our best option. Perhaps someone can comment on skylines ability/fitness for this. Peak quality score or to keep it absolutely basic, how about peak-width and peak symmetry (Asymmetry factor). I know you can export FWHM in a report, but can you get more custom LC characteristics or would the skyline framework be suitable for implementing this. Is it possible to add features to the list of possibilities in the Export-Report-options. Peak Width at 4-sigma, peak width at 10%, peak asymmetry at 4 sigma (or at half width). Don’t know if this would be in your definition of chromatographic quality, Brittany? Some examples here, Figure 1:

It could be useful for automatic QC (my focus is mainly LC). It is not that big a deal to write software for this from scratch, this is pretty basic stuff, but we need it for different vendors and different types of acquisitions, so I would like not to re-invent the wheel (again).

Best regards
bwidner responded:  2020-03-23
Yes, that is the kind of thing I am looking for!