research articles
Sweet et al.
Figure 2. Example of multiple precursor ions in isolation window. The identification was made by both Mascot and OMSSA after
removal of the entire precursor window. Unassigned peaks may be due to fragmentation of the unidentified coeluting precursor.
(Note that the noise peak has not been removed from this mass spectrum.)
Effect on Mascot score of removing precursor window
(precursor+/- 3 m∕z; 1628 DTAs)
140
120
100
DTA #
Figure 3. Effect on Mascot score of removing precursor window. A total of 3341 ECD DTAs (noise peak removed) were searched without
precursor window removal and with precursor window removal resulting in 1509 and 1609 identifications, respectively. A total of 1628
DTA were identified in one or both searches. The identifications are plotted, by ascending Mascot score of trimmed version (precursor
window removed) (red). The identification score for each DTA prior to trimming is shown alongside (blue). Reverse hits or rejected
hits (unacceptably large ppm error) were assigned a score of zero.
evenly into 10 bins, according to descending ECD Mascot
peptide score. The 21 conflicts and 16 reverse hits from the
initial search are shown separated across these bins, according
to the ECD Mascot score (Figure 6). Both conflicts and reverse
hits cluster around the lower Mascot scores, as would be
expected; however, even the lowest-scoring 10% of hits (Mascot
scores from 7.27 to 0.04) show high agreement with CID
identifications (96%) (6 conflicts) and an estimated FDR of only
6% (10 reverse hits).
To further validate these extremely low-scoring ECD iden-
tifications, the 83 identifications without a paired CID identi-
fication were examined manually. Characteristics of true
positive FT-ICR ECD identifications are the following: fragment
ion errors smaller than 12 ppm; y fragment ions, if present, of
5480 Journal of Proteome Research • Vol. 8, No. 12, 2009