ProjDIA: Fragment-Bin Projection Scoring and Calibrated Confidence for Direct Infusion DIA Proteomics

Abstract

Liquid chromatography–mass spectrometry (LC-MS) is the most powerful tool for bottom-up proteomics workflows but limits throughput because separations are slow and experimentally complex. Direct infusion shotgun proteome analysis (DI-SPA) offers a faster alternative, but highly chimeric tandem mass spectra complicate peptide identification and quantification. Here, we present ProjDIA, a computational workflow for DI-SPA that combines fragment-bin projection scoring, confidence-guided two-pass mass-error recalibration, monotone q-value reporting, and guarded semisupervised rescoring. Across public HeLa DI-SPA benchmarks, ProjDIA increased peptide-level identifications relative to CsoDIAq and showed condition-dependent gains relative to CHIMERYS, with the clearest advantage observed in the sample-load series and at higher MS2 resolution. In species-entrapment analysis, ProjDIA showed that these gains are not at the expense of higher error rates. Finally, ProjDIA also supported label-free quantification and demonstrated the gains in peptide identification can lead to deeper coverage of proteins with significant fold changes under perturbation.