Who wants to manually quality check peptides?!

Took 8.5 hours on a Sunday night, but I calculated an error rate for Skyline! I randomly selected 100 proteins in Skyline, and for each protein I checked one peptide. Ideally, these peptides had data availabe for every replicate. I evaluated whether or not Skyline chose the right peak, and if the peak boundaries were correctly deliniated.

For each peak, I assigned a “0” if Skyline picked the wrong peak and a “1” if it picked the peak correctly. For some peptides, there was no data available for a replicate. This would be a blank screen, so I gave it the value “N/A.” I then calculated error rates per replicate and peptide. You can find my spreadsheet here.

Here are some examples of Skyline spectra.

1: Skyline picked the right peak

0: I assigned this value when Skyline picked the wrong peak. I also assigned this value when Skyline picked a peak where there was just noise.

Wrong peak:

Noise:

Changing peak boundaries: