Is it possible to get mProphet scores at the precursor level rather than the peptide level?

support
Is it possible to get mProphet scores at the precursor level rather than the peptide level? brian24641  2022-01-22 12:18
 

In the mProphet paper it looks like they generate scores at the precursor level i.e. one score for each potential peptide charge (typically +2 and +3). However when I look at the exported mProphet features they appear to be at the peptide level. When I then generate a report with Detection Q Values for each precursor I see that all precursors associated with a peptide for a replicate have the same value, which implies that mProphet results were rolled up to the peptide level from the precursor level implicitly. Thus, is it possible to get these scores at the precursor level, and also what the algorithm is that is being used to roll up mProphet scores (or Q values) to the peptide level?

 
 
Nick Shulman responded:  2022-01-22 14:21
mProphet scores are usually calculated at the peptide level.
If you applied an isotope modification to the heavy peptide, and that modification had the "Relative retention time" set to "Unknown", then Skyline would calculate the mProphet scores separately for the light and heavy precursors. But, even in that case, the charge states of each label type would be considered together.

So, typically, all precursors under a peptide will have exactly the same mProphet scores. Those scores appear at the precursor level in the Document Grid because sometimes, depending on the isotope modifications you have in your document, the light and heavy scores might be different from each other.

If you wanted to get separate scores for your different charge states, one thing that you could do is delete everything but the charge 2 precursors from you document, and then do "Edit > Manage Results > Rescore" and then "Refine > Reintegrate" and see what scores all of the peptides got.

When Skyline is doing peak picking, Skyline always looks at the transitions for all of the charge states at the same time.
-- Nick