Answers are grounded in this paper and cite their sources.
Data-independent acquisition (DIA)-mass spectrometry (MS) data is typically searched using spectral reference libraries (SRLs) created from data-dependent acquisition (DDA). Shortcomings with this approach include inherent missingness, low reproducibility, and bias to high abundant peptides. We systematically compared the performance of four SRL generation strategies: (1) fractionated DDA, (2) fractionated DIA, (3) in silico SRL derived library refined using unfractionated sample DIA files (SRL-free DIA), and (4) gas phase fractionation (GPF) DIA on 9 cancer cohorts accounting for 246 human samples. Each was evaluated in terms of library size, quantitated peptides, q-value scoring, missingness, peptide intensity, and differentially abundant (DA) proteins. In a prostate adenocarcinoma set, fractionated DDA displayed the highest number of identifications, while SRL-free DIA had the lowest number of missing values (∼40%). Peptides uniquely quantified by SRL-free DIA showed favorable within-search scoring characteristics, higher intensity values (>50% improvement), and the lowest missingness (14% improvement) leading to more significantly reported DA proteins. Fractionated DDA SRLs performed less favorably in most of the metrics tested, challenging the assumption that larger SRLs improve high-quality quantitation. SRL-free DIA closely represents the samples under study and offers a higher quality alternative for large-scale clinical proteomic analysis where sample availability is limited.
No takes yet. Share an insight, caveat, or question.
Bucio-Noble et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: