spectrum matched to a wrong peptide

[treko90]

Hi,

I am trying to decide whether my spectrum (https://limewire.com/d/WfWzE#GsF4YEsPHf) corresponds to peptide STAIGNLLEPR (novel variant) or STANGNLLEPR (reference proteome variant). PepQuery confirms that spectrum corresponds to novel variant (I on position 4 and not N). However, there are multiple fragments confirming N at position 4 and not I:

y8⁺ (N-variant): theo 912.4898, obs 912.4882 (−1.7 ppm)
y9⁺ (N-variant): theo 983.5269, obs 983.5298 (+3.0 ppm)
b4⁺ (N-variant): theo 374.1670, obs 374.1675 (+1.3 ppm)
b5⁺ (N-variant): theo 431.1885, obs 431.1872 (−3.0 ppm)
b7⁺ (N-variant): theo 658.3155, obs 658.3150 (−0.8 ppm)
b9²⁺ (N-variant): theo 450.2211, obs 450.2236 (+5.6 ppm)
b8²⁺ (+1 iso) (N-variant): theo 386.2058, obs 386.2058 (+0.0 ppm)

How come PepQuery rejects the reference variant even though there is evidence for N4 instead of I4?

My fasta file is here:
https://limewire.com/d/anZVy#xUxxFyeRwz

spectrum comes from deep 29 tissues dataset, commands I used for validation:
java -jar pepquery-2.0.2.jar -db protein.fasta -ms spectrum.mgf -ti 0,1,2,3 -s 2 -fast -fixMod 1 -varMod 2 -tol 20 -itol 0.05 -i STANGNLLEPR
java -jar pepquery-2.0.2.jar -db protein.fasta -ms spectrum.mgf -ti 0,1,2,3 -s 1 -fast -fixMod 1 -varMod 2 -tol 20 -itol 0.05 -i STAIGNLLEPR

[treko90]

I realized that precursor mass did not correspond to N variant even though fragment ions support N variant better than I.
In result PepQuery did not continue to evaluate MS2 fragments. Maybe it is worth to implement MS2-only validation so the scores for different hypotheses can be compared in case the spectrum is chimeric or MS1 is flawed somehow. Now i worked around it by putting the expected precursor mass in the mgf spectrum. When I run validation of this modified mgf and the original one I get two scores corresponding to reference and novel variant which I can compare

[treko90]

Also, matching only MS2 level evidence ignoring precursor mass may help with discrimination of common labile modifications. Currently, when validating a novel peptide, pepquery may falsely match a spectrum to the novel sequence on the basis of precursor mass. However, precursor may be a reference peptide with labile modification (such as phsphorylation). If we test for a single amino acid variant, similarity of a novel peptide with the reference variant may further confirm the validity even though matching MS2 to the reference peptide (which is not considered for competitive scoring now) would yield a higher score indicating the loss of PTM during fragmentation

[wenbostar]

Also, matching only MS2 level evidence ignoring precursor mass may help with discrimination of common labile modifications.

Without the precursor mass constrain, it seems it's hard to evaluate which match is more confident.