Question about calculating standard errors of mean and CVs for group comparison

Question about calculating standard errors of mean and CVs for group comparison roman sakson  2020-09-06

Dear Skyline team,

I did an experiment where I monitored four peptides for a particular protein and afterwards decided to include all transitions for my first peptide but only one quantitative transition for the other three peptides, respectively, when performing the group comparison. I noticed that the peptide with several transitions had a higher significance in the volcano plot compared to others (see volcano_plot.jpg) and I investigated further. When I had a look at the respective CVs for all transitions for that peptide across the two conditions, CV values were terrible (around 90 %, see CVs_Transitions.jpg), which did not agree with the significance in the volcano plot nor with my expectations based on "okayish" to good XICs. However, once I summarized transitions from "all" to "total", overall CVs dropped below 10 %, which would correspond to the volcano plot (see CVs_All.jpg). For the other peptides, where there are no transitions to sum up, I do not see this pattern. This effect is present in the latest daily version as well as in 20.1. I assume that CVs for individual transitions are somehow off. Do you have an idea what the reason might be?

Thank you,


Nick Shulman responded:  2020-09-06

The reason that the CVs are so high in your CVs_Transitions.jpg is that in each of those two groups, there is one replicate whose value is zero.
I am not sure what is going on with those zero replicates. When I look at their chromatogram those replicates show "Chromatogram information unavailable".

The inclusion of these zero values in the CV calculation in the Peak Area Replicate Comparison window causes the CV to be a high number.
These zero values do not end up getting included in the Group Comparison calculation.

I am not sure exactly why the peak area graph decided to include these two replicates, and the group comparison does not include them.
It might be that the more relevant question is why your data looks the way it does. Do you know what is going on with the replicates U_POMT1_III and ALG1_I?
-- Nick
roman sakson responded:  2020-09-07

thank you for your answer. This is true, for the two replicates U_POMT1_III and ALG1_1 the peptide in question was not included in the transition list, I added it later on. So, there is simply no data available. From my understanding for data analysis, it is correct not to include these zero values in the Group Comparison calculation and instead to influence the statistical outcome by substracting 2 degrees of freedom, which, I assume, Skyline does correctly.

I guess, what made me ask in the Support Board was the fact that the Peak Area Replicate Comparison plot itself treats these values differently, depending on whether Transitions are set to "All" or to "Total" and I was not sure, whether this is a bug.

Thank you,