Properties of Average Score Distributions of SEQUEST

• DOI: 10.1074/mcp.m700239-mcp200
• Publication Date: 2008-02-27T01:14:37Z
• Authors: Salvador Martiáne...; Pedro Navarro; Fernando Martián-...; Daniel López-Ferrer; Antonio Ramos-Fer...; Margarita Villar; Josefa P. Garciáa...; Jesuás Vázquez
• Keywords: Statistical power

High throughput identification of peptides in databases from tandem mass spectrometry data is a key technique modern proteomics. Common approaches to interpret large scale peptide results are based on the statistical analysis average score distributions, which constructed set best scores produced by collections MS/MS spectra using searching engines such as SEQUEST. Other calculate individual probabilities basis theoretical models or single-spectrum distributions each spectrum. In this work, we study mathematical properties SEQUEST introducing concept spectrum quality and expressing these compositions distributions. We predict demonstrate practice that dominated distribution collection, except low probability region, where it possible dependence database size. Our leads novel indicator, ratio, takes optimally into account information provided first second scores. The ratio non-parametric robust indicator makes classification according parameters charge state unnecessary allows performance, false discovery rates, better than obtained other empirical approaches. also compares favorably with indicators construction These make robustness, conceptual simplicity, ease automation algorithm very attractive alternative determine confidences error rates high experiments.