iRT on non linear gradient

support
iRT on non linear gradient benoit fatou  2021-01-04
 

Dear Skyline team,
I was wondering how I can use the iRT peptides for retention time adjustment when using a non-linear gradient.
Thanks for your help and Happy New Year !
Benoit

 
 
Nick Shulman responded:  2021-01-04
You need to give all of your peptides and standards new iRT values.

What I would recommend doing is removing the retention time predictor from your Skyline document, and then extracting chromatograms from a few replicates. Manually make sure that Skyline has chosen the correct peak for all of your peptides.

Then, go to:
Settings > Peptide Settings > Prediction > Calculator Button > Add
and create a new iRT database.
You will need to use the "Calibrate" button in order to assign appropriate iRT values to standards based on where those peptides were detected.

-- Nick
 
benoit fatou responded:  2021-01-04
Thanks for your reply Nick.
I have a question about what you are suggesting: after the appropriate iRT values edited, how does Skyline know that the retention time predictor will use the non-linear gradient settings in case of RT realignment?

Also, in which case(s) should we use the logarithmic or loess regression types?
Thanks for your help,
Benoit
 
Nick Shulman responded:  2021-01-04
Skyline does not care whether the gradient is linear or some weird curve. All that Skyline cares about is that there is a linear relationship between the iRT values for the peptide and the retention time at which the peptide elutes.

Even if your instrument method involves changing the mixture of solvents at a non-linear rate, it is always possible to find a set of iRT values for the peptides such that the mapping between iRT value and retention time is linear.

Is there something that I am not understanding about your question?

I am not sure whether there is ever any real reason to use the "Logarithmic" regression type in the Edit iRT Calculator dialog. That option was added because there was a scientist who had a set of iRT standards where there was a logarithmic relationship between the molecular weight of the molecule and its retention time, and, this option allowed the molecular weight to be used as the iRT score.

The Lowess option is good if you have a large number (a few dozen, maybe a hundred; I'm not sure what we intended) iRT standards. When you choose Lowess, instead of Skyline drawing a straight line between the iRT standards, you get a winding curve which in theory will much more precisely predict the retention time of all of the other peptides.
-- Nick
 
benoit fatou responded:  2021-01-04
Thanks! That is very helpful.

According to you just said, it might not be possible to refine the retention time from a linear to a non-linear gradient by "just" doing an unscheduled MRM experiment of the iRT peptides.
Thoughts?
Thnaks,
Benoit
 
Nick Shulman responded:  2021-01-04
Yes, I think that is possible.

On the Edit iRT Calculator dialog, if you push the "Add > Add iRT Database" button, I think Skyline transforms the iRT values of the new peptides using a linear regression on the peptides that are common between the two databases.

Someone else might be more of an expert on how this works. I am not sure whether that linear regression is performed on just the iRT standards, or whether all the common peptides participate. I also don't know whether the "Regression Type" on the Edit iRT Calculator dialog has any impact when you are adding another iRT database.
-- Nick
 
benoit fatou responded:  2021-01-04
I will definitely try that this week and get back to you if I have any questions/comments.

Thanks for your help,
Benoit