Nature of False Peptide Identifications in Data-Independent Acquisition-Based Proteome Analysis

Data-independent acquisition (DIA) mass spectrometry is increasingly used in proteomics because it offers a shorter analysis time at higher proteome coverage. Most existing DIA search engines rely heavily on fragmentation spectra and produce identification results, even when precursor ions are not detected. Although the latter is well-known, the errors it can lead to have received little attention. In this work, we studied false identifications in DIA, primarily due to the lack of a corresponding precursor ion envelope. Using a DIA data set with known UPS proteins spiked into Escherichia coli (E. coli), we performed an extensive search against several hundred databases with in silico-generated UPS variants mimicking single amino-acid substitutions. We found that DIA search results often fail to distinguish between peptides with similar fragmentation patterns but different precursor masses. Peptides with substitutions near the N-terminus were more often misidentified by the search engine than those with substitutions near the C-terminus. This algorithmic shortcoming limits its applications in areas such as proteogenomics, post-translational modification detection, and proteoform analysis. To partially address this issue, we propose using identifications based only on the identified precursor and the narrow isolation windows.