Evaluating Mascot version and percolator-based rescoring on protein identification confidence, with a Mascot Distiller quantitation pipeline demonstration

ABSTRACT: We compared two versions of the Mascot search engine (2.6.2 and 3.1) and evaluated Mascot’s Percolator-based machine-learning rescoring (ML-boost) on HeLa and CHO LC-MS/MS datasets. The version change had little effect on identifications, while ML-boost substantially increased protein numbers, by about 46% in CHO and 9% in HeLa. However, most proteins gained only through ML-boost were low-confidence, single-peptide identifications with markedly lower scores. These results suggest that rescoring gains should be interpreted with caution, especially when they come from single-peptide evidence.

ORGANISM(S): Cricetulus Griseus

SUBMITTER: Arman Kulyyassov  

PROVIDER: PXD085963 | iProX | Tue Oct 20 00:00:00 BST 2026

REPOSITORIES: iProX