We agree that many low confidence peptides are included in our analysis. We decided to do that for the following reasons:
1) We were looking at the data collectively, to analyze trends of peptide abundance in two samples (GILT-/- and WT), rather than individual sequence of any particular peptide. Keeping the confidence level very high might improve the quality of the data, but ultimately, it may also skew the analysis by keeping only highly abundant peptides.
2) We found same peptide(s) in two independently isolated samples (GILT-/- and WT) labeled by iTRAQ114 and iTRAQ117, using multiple mass data analysis with different protocols. The probability of this happening due to a “poor quality MS/MS data” is not very high, poor quality of data would more likely result in randomly different samples.
3) The appearance of numerous low confidence peptides is an inherent problem with the samples of high complexity and low individual abundance. We undertook the best techniques with a large number of mice > 100, improved purification/ labeling protocol inclining multi-dimensional chromatography, and used the most sensitive instrument to improve the detection sensitivity.
4) Many peptides with low confidence levels have been analyzed manually looking back into the raw data, paying particular attention to the fragmentation and to alternative peptide sequences that may fit to the fragmentation patterns.
5) The ability of some of the peptides to bind to MHC class II was validated experimentally.

Your second concern was finding of “exotic post-translational modifications that were previously not found in peptides bound to MHC molecules”. While we agree, and have actually pointed out in the manuscript, that some of these modifications might have been introduced during the sample preparation, we strongly disagree with the implied reasoning (since these modifications have not been reported previously, they must be artifactual). Studying post-translational modifications of proteins and peptides at the global level has become possible only recently and not much is known yet or established. A large number of proteins are post-translationally modified and there is no reason to suggest that their degradation products (which MHC bound peptides are) are systematically depleted of modifications prior to or after binding to MHC molecules. These modifications may, or may not play important roles, but either way we should not ignore them.