A note on the false discovery rate of novel peptides in proteogenomics

Abstract

Motivation: Proteogenomics has been well accepted as a tool to discover novel genes. In most conventional proteogenomic studies, a global false discovery rate is used to filter out false positives for identifying credible novel peptides. However, it has been found that the actual level of false positives in novel peptides is often out of control and behaves differently for different genomes.Results: To quantitatively model this problem, we theoretically analyze the subgroup false discovery rates of annotated and novel peptides. Our analysis shows that the annotation completeness ratio of a genome is the dominant factor influencing the subgroup FDR of novel peptides. Experimental results on two real datasets of Escherichia coli and Mycobacterium tuberculosis support our conjecture.Contact: yfu@amss.ac.cn or xupingghy@gmail.com or smhe@ict.ac.cnSupplementary information: Supplementary data are available at Bioinformatics online.

GAPP: A Proteogenomic Software for Genome Annotation and Global Profiling of Post-translational Modifications in Prokaryotes

“…To accurately identify novel peptides, it is necessary to filter the search results to control the FDR. Recently, more stringent filtering strategies, such as posterror probability (89) or separate FDRs for annotated and novel peptides (46,90) were employed in proteomic analysis. In this work, we used a more stringent strategy (novel FDR) to increase the accuracy of identified novel peptides (76,77).…”

“…In this work, we used a more stringent strategy (novel FDR) to increase the accuracy of identified novel peptides (76,77). However, because of the lower expression levels or worse scores of the novel peptides in proteogenomic study, the FDR calculation is difficult to explicitly estimate to exclude false peptides (90). To further confirm the identified novel genes, GAPP applied a postprocessing of identification strategy, including the novel protein sequence alignment and GO functional annotation.…”

“…The search space inflates dramatically with each additional variable modification added in the "restricted search" mode, leading to low interpretation rates of MS/MS spectra and high false-positive rates (93,94). Recently, the unrestricted database search algorithms, which do not require prior knowledge of the modifications present in the sample, are used in high-throughput proteome analysis (90). GAPP integrated the unrestrictive PTMs search strategy (MODa) into the pipeline (64) and identified a large number of novel modifications in H. pylori.…”

Connecting Proteomics to Next‐Generation Sequencing: Proteogenomics and Its Current Applications in Biology

“…Zhang et al. investigated both global and separate FDR settings respectively for sub‐groups of annotated and novel peptides and demonstrated that the annotation completeness ratio of a genome is the dominant factor influencing the subgroup FDR of novel peptides . Another popular method (such as the Percolator) to estimate FDR is by coupling semi‐supervised learning to a decoy database search strategy so as to discriminate correct and false PSMs; often with extra information (e.g., retention time of chromatography, peptide charge state, or mass accuracy) from MS/MS experiments to re‐score PSMs .…”

A tool for integrating genetic and mass spectrometry‐based peptide data: Proteogenomics Viewer

“…In contrast, proteogenomics studies investigating alternative splicing or noncoding regions require custom made databases containing "novel" sequences predicted either from genome or transcriptome (ESTs, RNASeq, etc.) However, the identification of such novel peptide variants in proteogenomics studies are still susceptive to higher than expected falsediscovery rates (FDR) [8,9], mostly due to the larger search space of the customized databases (and consequently higher chance of spurious matching). However, the identification of such novel peptide variants in proteogenomics studies are still susceptive to higher than expected falsediscovery rates (FDR) [8,9], mostly due to the larger search space of the customized databases (and consequently higher chance of spurious matching).…”