Fig. 2: Evaluating empirically corrected libraries made with peptide predictions.

a Violin plots showing spectral correlation between predicted library for yeast peptides and single-injection DIA (N = 1) at various NCE settings (circles indicate medians) when the instrument was configured for NCE = 33. b The total numbers of empirically corrected library entries detected from GPF-DIA at various NCE settings (N = 1). c The fraction of peptides detected in single-injection DIA (N = 1) relative to the optimal NCE for empirically corrected libraries (blue line) are more consistent than predicted libraries (red dashed line) across a wide range of NCE settings. d The number of yeast peptides detected at 1% peptide FDR in single-injection DIA acquisitions (N = 4) using the NCE = 33 chromatogram library, where either the retention times or fragmentation patterns have been switched with the predicted Prosit values (purple bars). Compared to the predicted spectrum library search (red bar), an 11% increase comes from simply using a narrowed peptide selection. DIA-based retention times and fragmentation patterns provide a 13% and 10% increase over this, respectively. Comparing the empirically corrected library and the predicted library detections (blue bar, 37% increase), these percentage gains appear to be nearly multiplicative (i.e., 111% × 110% × 113% = 138%), indicating that all three factors are independent and of roughly equal importance. Source data are provided as a Source Data file.