Hello,

Deciding how to deal with peptide/protein mapping and its ambiguities is a tricky problem.

If you want to be doing spectra counting at the protein level, you can choose to do your comparison based on the protein assigned by either XTandem or by ProteinProphet, instead of rolling up the counts based on the individual peptide sequences. However, this will hide information about the relative frequency of the different peptide identifications themselves.

Alternatively, you could group the counts by both the peptide sequence and protein assignment, and then do your own analysis in an external tool (like R) based on your own criteria.

I realize this isn't a terribly specific answer, but I hope it's still useful.

Thanks,
Josh