Error in Accuracy Calculation when Requiring Sample Dilution Factor Will Thompson  2021-04-07 07:44
 

Dear Skyline friends. Reporting a bug in Skyline calibration, wherein the accuracy is calculated incorrectly. Specifically, if linear regression against any external calibration data is performed and a 'sample dilution factor' is required for a quality control sample which may use a different volume than the standard curve, the calculated concentration is updated correctly, but the accuracy is not correct. This causes problems when a sample or QC volume (for instance) is different than a calibrator volume, and a sample dilution factor is needed. This hasn't been noticed before on my end because accuracy isn't usually calculated for unknowns, however i recently had a case where a different prep volume was needed for some samples in a batch, and therefore i prepared an additional set of quality control samples using the same approach as the samples needing dilution, to make sure the backcalculated concentration was accurate using the different sample volume. In the skyline document attached, you will see four QC runs which are called "QC1_45uL_01" (and others with 45 uL in the name) where (for example) the target concentration is 1 nM, the calculated concentration is 1.006 nM, but the accuracy is calculated as 110.7%. A sample dilution factor of 1.1 was used. Accuracy is calculated correctly for all other QCs which did not require a sample dilution factor. It seems like the accuracy calculation within skyline may be using some method other than comparing the "calculated concentration" field to the "Analyte Concentration*Concentration Multiplier" (which seems like would be the correct approach)

Thanks for looking into this.

Cheers

W

 
 
Nick Shulman responded:  2021-04-07 09:15
Thanks for pointing this out. I will fix it in an upcoming release of Skyline-Daily.

-- Nick