DIA extraction window

support
DIA extraction window joanna bons  2017-08-21 02:36
 
Hello,

I am analyzing DIA data using Skyline.
iRT peptides have been spiked into the samples for retention time calibration. I have reintegrated peaks for some iRT precursors, which resulted in a new regression. Afterwards, I have re-imported the results.
But, for some peptides of interest, the extraction window is not centered around the predicted RT, as it is specified in the Transition Settings".
Should I thus delete the iRT peptides which are not well integrated from the Calculator +/- the target list?

Thank you for your help,
Joanna Bons.
 
 
Brendan MacLean responded:  2017-08-21 10:45
At present, there is not a way to inform Skyline manually of the true locations of the iRT peptides during the results import. The retention time ranges for extraction will be based on the Skyline automated detection of the iRT peptides in the extracted full-gradient chromatograms.

Skyline does use an algorithm for discarding outlier peaks from among your iRT standard peptides, but it is still all automated with no ability to correct Skyline, even if you correct the peak integration and then use Edit > Manage Results and the Re-import button. Though, that sounds like a perfectly fine idea. So, maybe we will look into supporting it in the future.

At the moment, however, you are stuck with the fully-automated algorithm. So, the only way of improving its performance would be, as you suggest, improve the reliability with which Skyline detects your iRT standard peptides by removing the weaker performers and potentially adding new more reliably detected peptides. For example, I have heard that on some chromatographic systems the first eluting peptide in the Biognosys kit is problematic for people, and so only the 10 later-eluting peptides are used.

Despite the automatic attempts of Skyline to remove outliers, it is better to have no outliers and instead have only stable, reliably detected peaks. Problems with the peptides near the ends are also overweighted in the regression, compared to peptides in the middle. So, be especially careful that your standards near the end are reliable.

Hope this helps. Sorry, we do not yet support the manual integration correction you attempted. Thanks for posting to the Skyline support board.

--Brendan
 
joanna bons responded:  2017-08-22 05:41
Hello Brendan,

Thank you very much for your response.

I have just one more question regarding the regression method.
By cliking right on the graph representing the measured times over the calculator, it is possible to change the regression method from linear regression to Kernel Density Estimation or LOESS. Does choosing here Kernel Density Estimation, LOESS or linear regression will have an influence on the predicted retention time? Or the regression equation is still the same (a linear one) as the one defined by the Calculator, and the choice of the regression method will just have an "impact" on the graph mentionned previously?

Thank you,
Best regards,
Joanna.
 
Brendan MacLean responded:  2017-09-10 17:56
These are at present just visualizations that we are using to assess the benefit of different types of regression, and you can too, but we are not yet using them for prediction, except on a specific development branch with is not yet released.

Thanks for your interest.

--Brendan