• Sign In
  1. Requests

background height

support

View Request

view list print
background height rameshkr  2025-03-01 12:21
 

Hi team

I want to use the RMS method for signal to noise ratio calculation using peak heights of individual peptides.
Where can i get the background height data? And should i combine the background height for each transition/precursor to get the SNR for a given peptide? Or is there a better way to do the RMS based SNR calculation?

Thanks
Ramesh
 
 
Mike MacCoss responded:  2025-03-01 19:14

Hi Ramesh,
Thanks for your interest. We've not reported a signal-to-noise calculation in Skyline partially because it isn't straight forward and different people think about S/N differently. Also, depending on the instrument platform and the degree of preprocessing the S/N can be significantly distorted from reality. For example, Orbitrap analyzers remove the background as a type of lossy signal compression. This compression makes a lot of practical sense but it means that any type of S/N calculation performed on the RAW data is going to appear unrealistically good. Furthermore, most of the "noise" is signal dependent and "shot noise" isn't taken into account when using a measure of background for noise. I frequently suggest people use 1/CV from replicates when people want a rough estimate of the (mean background subtracted signal)/(noise in the signal).

That said, over the years we have discussed that it might be useful to have some measure of S/N so that a user can say if the signal is statistically different from the background. I personally believe we need to use the background subtracted peak height as the signal -- this is currently implemented. However, for the "noise", we don't currently have a measure of the variance of the background yet. I've prefered using the standard deviation of the background as the noise as it provides a spread of the background. However, where the "noise" is sampled can significantly skew the estimate of the noise -- especially if the noise is being estimated without a matrix blank. Furthermore, this measurement would still be heavily biased for some instruments where there is significant processing prior to writing the raw data.

Anyway, we have implemented S/N into some new search tools and it is possible to get the background subtracted peak area (signal) and the background out of our current data grid in Skyline. I hope we have something soon but I am hesitant as it can be difficult to interpret and easy to get a value that can be misinterpreted.

Sorry I don't have a better answer,
Mike