Hi Mathilde,
Thanks for your email. The problem is what do you consider your noise? Are you looking for the background? Skyline can output the background in the grid or a report. There is Total Background, Total Background MS1, and Total Background Fragment. For signal you can just use the peak area, which is the background subtracted signal. So you can use the grid to output a report of signal and background fairly easily to report the signal/background easily.

In a quantitative experiment we tend to think of noise as variance. In this case the mean-signal/stdev would be better measure of S/N. This of course is the reciprocal of CV which is a common figure of merit used in most quantitative experiments.

I hope this helps. We haven't reported signal to noise specifically because "noise" is not something that is well standardized. In our lab we just use CV as that generally answers what we are most interested in -- how reproducible are our measurements.

Cheers,
Mike

Hi Mike,

Thanks a lot for your answer; In its publication, Catenacci calculate a signal to noise ratio to determinate LOD (limit of detection) and LOQ (limit of quantification) for absolute quantification. To calculate this S/N ratio, they add decoy peptides before analyzing their row files, and make a peak scoring model using mProphet. But they work on SRM. I try to calculate this S/N ratio with PRM, unfortunately, I can not integrate decoy peptides in the analyses of my row files. Therefore I can't calculate this ratio.

Thanks for your help!

Mathilde

Do you mind sending the reference? Within Skyline we have mProphet implemented and we also have worked on methods to compute LOD and LOQ. However, I am not sure how those are relate to S/N directly. I'm not sure what is being used as the "noise." I can imagine that someone could use mProphet to define what peptides are above the LOD. However, I am not sure why one would need the S/N for that. Does the approach require that you inject a blank to assess the noise?

It would be great to understand this more.

-Mike

Find the publication attached. I highlighted the determination of LOD and LOQ in the method section, generation of standard curve. Find also the tuto of peak picking model to calculate S/N ratio

Mathilde

I think they must be using a signal to background measurement for this. You should be able to estimate the background as I describe above. If you have a better measure for noise let us know.
Cheers,
Mike