vdemichev
commented
1 month ago Hi,
A 'decoy' channel contains in silico generated precursor ions with the label masses as DIA-NN specifies in the log. Since any peptide-spectrum match in this channel is known to be false, it allows to estimate the numbers of precursors with false channel assignment in the other channels, and hence calculate channel q-values. These channel decoys do not participate in the calculation of the regular FDR, i.e. this is purely for evaluating channel confidence.
But with these additional decoys, maybe there is no such problem?
Exactly!
Best, Vadim
I am applying SILAC modifications to a spectral library using these options:
--fixed-mod SILAC,0.0,KR,label --lib-fixed-mod SILAC --channels SILAC,L,KR,0:0; SILAC,H,KR,8.014199:10.008269 --original-mods
Interestingly, in the log I see the following:
Registered SILAC channel L with following masses: K[0] R[0] Registered SILAC channel H with following masses: K[8.0142] R[10.0083] Registered SILAC channel decoy with following masses: K[16.0284] R[20.0165]
What are the channel decoys? Is this an additional set of decoys which have the same sequence as the targets, but a different mass offset for the labeled amino acids? So then there would be two decoys for each target, one shuffled and one labeled "extra" heavy.
I assume this is to avoid falsely attributing labeled library peptides to spectra produced by unlabeled peptides and vice-versa. For example if I searched an unlabeled sample with labeled sequences added to the library I would probably identify heavy labeled peptides at a rate higher than my FDR threshold? But with these additional decoys, maybe there is no such problem?
Thanks!