Spectral prediction features as a solution for the search space size problem in proteogenomics

Verbruggen et al. (2021) address a core challenge in proteogenomics: as Ribo-seq and Nanopore RNA-seq expand the protein sequence search space, distinguishing true from false peptide-to-spectrum matches becomes increasingly difficult. The authors demonstrate that spectral intensity prediction features, extracted from the MS2PIP and Prosit predictors and integrated with canonical MaxQuant scores via the Percolator post-processing tool, substantially improve peptide identification confidence in large, Ribo-seq-derived databases. Applied to search spaces built from combined ribosome profiling and Nanopore long-read sequencing data, this approach increases sensitivity without compromising specificity, offering a practical solution for high-confidence novel proteoform discovery.