Statistical Validation of Peptide Identifications in Large-Scale Proteomics Using the Target-Decoy Database Search Strategy and Flexible Mixture Modeling

Choi, Hyungwon; Ghosh, Debashis; Nesvizhskii, Alexey I.

doi:10.1021/pr7006818

Cited by 112 publications

(163 citation statements)

References 26 publications

Supporting

Mentioning

161

Contrasting

Order By: Relevance

“…The whole dataset of the peptide identifications in practice is usually sufficiently large that FDR(x) can always be accurately estimated by using the target-decoy strategy or more complicated approaches (24,25). However, the number of identifications of k-modified peptides could be too small to support a separate, accurate estimation of FDR k (x).…”

Section: And Methodsmentioning

confidence: 99%

Transferred Subgroup False Discovery Rate for Rare Post-translational Modifications Detected by Mass Spectrometry

Qian

2014

Molecular & Cellular Proteomics

122

View full text Add to dashboard Cite

In shotgun proteomics, high-throughput mass spectrometry experiments and the subsequent data analysis produce thousands to millions of hypothetical peptide identifications. The common way to estimate the false discovery rate (FDR) of peptide identifications is the target-decoy database search strategy, which is efficient and accurate for large datasets. However, the legitimacy of the target-decoy strategy for protein-modification-centric studies has rarely been rigorously validated. It is often the case that a global FDR is estimated for all peptide identifications including both modified and unmodified peptides, but that only a subgroup of identifications with a certain type of modification is focused on. As revealed recently, the subgroup FDR of modified peptide identifications can differ dramatically from the global FDR at the same score threshold, and thus the former, when it is of interest, should be separately estimated. However, rare modifications often result in a very small number of modified peptide identifications, which makes the direct separate FDR estimation inaccurate because of the inadequate sample size. This paper presents a method called the transferred FDR for accurately estimating the FDR of an arbitrary number of modified peptide identifications. Through flexible use of the empirical data from a targetdecoy database search, a theoretical relationship between the subgroup FDR and the global FDR is made computable. Through this relationship, the subgroup FDR can be predicted from the global FDR, allowing one to avoid an inaccurate direct estimation from a limited amount of data. The effectiveness of the method is demonstrated with both simulated and real mass spectra. Molecular & Cellular

show abstract

Section: And Methodsmentioning

confidence: 99%

Transferred Subgroup False Discovery Rate for Rare Post-translational Modifications Detected by Mass Spectrometry

Qian

2014

Molecular & Cellular Proteomics

122

View full text Add to dashboard Cite

show abstract

“…This is a different and less powerful classification rule than PeptideProphet would ordinarily compute, which uses densities and results in spectra specific error rates, but we use it here in order to compare this with the decoy database threshold-based approaches. See equation (7) on page 11 of Choi and Nesvizhskii [2] for a more general calculation. When searched against a decoy database, this PeptideProphet FIR may be estimated empirically by (# decoy > x)/(# target > x), which is equal to the classic target FIR described above.…”

Section: Perspectivementioning

confidence: 99%

Modes of Inference for Evaluating the Confidence of Peptide Identifications

Fitzgibbon

McIntosh

2007

J. Proteome Res.

View full text Add to dashboard Cite

Several modes of inference are currently used in practice to evaluate the confidence of putative peptide identifications resulting from database scoring algorithms such as Mascot, SEQUEST or X!Tandem. The approaches include parametric methods, such as classic PeptideProphet, and distribution free methods, such as methods based on reverse or decoy databases. Due to its parametric nature classic PeptideProphet, although highly robust, was not highly flexible and was difficult to apply to new search algorithms or classification scores. While commonly applied, the decoy approach has not yet been fully formalized and standardized. And, although they are distribution-free, they like other approaches are not free of assumptions. Recent manuscripts by Kall et al, Choi and Nesvizhskii and Choi et al help advance these methods, specifically by formalizing an alternative formulation of decoy databases approaches and extending the PeptideProphet methods to make explicit use of decoy databases, respectively. Taken together with standardized decoy database methods, and expectation scores computed by search engines like Tandem, there exist at least four different modes of inference used to assign confidence levels to individual peptides or groups of peptides. We overview and compare the assumptions of each of these approach and summarize some interpretation issues. We also discuss some suggestions, which may make the use of decoy databases more computationally efficient in practice. PerspectiveInterpreting tandem mass spectrometry (MS/MS) experiments involves a large number of computational steps, the first of which is to assign to each of thousands of spectra a single putative sequence and an associated score related to the accuracy of the assignment. The second step is that of assigning an interpretable measure of confidence to one or a group of those identifications. Because all subsequent results depend highly on these confidence measures this step may be the most fundamental of all. Several different modes of inference are now used to assign confidence measures, each of which relies on the same underlying model yet proceeds using a different class of assumptions. In our discussion we will follow the language suggested by Kall et al [1] with one exception: rather than the term False Discovery Rate (FDR) we suggest the use of False Identification Rate (FIR). Although we recognize that they are derived from the same underlying statistical foundations it is

show abstract

“…The search included fixed modification of carbamidomethylation on cysteine, and variable modifications: oxidation (Met), Gln 3 pyro-Glu (N-terminal Gln), Glu 3 pyro-Glu (N-terminal Glu), and deamidation (O 16 ) on both asparagine and glutamine, and O 18 -incorporated deamidation on asparagine. The FDR analysis used the R implementation of a flexible mixture model as described by Choi et al to calculate global and local peptide-level false discovery rates (23). The false discovery rate (FDR) was controlled at 1% at the peptide level by searching the same data set against a decoy database, and a minimum of two unique peptides per protein was required for the identification of proteins.…”

Section: Methodsmentioning

confidence: 99%

The GlycoFilter: A Simple and Comprehensive Sample Preparation Platform for Proteomics, N-Glycomics and Glycosylation Site Assignment

Zhou

Froehlich

Briscoe

et al. 2013

Molecular & Cellular Proteomics

View full text Add to dashboard Cite

Current strategies to study N-glycoproteins in complex samples are often discrete, focusing on either N-glycans or N-glycosites enriched by sugar-based techniques. In this study we report a simple and rapid sample preparation platform, the GlycoFilter, which allows a comprehensive characterization of N-glycans, N-glycosites, and proteins in a single workflow. Both PNGase F catalyzed de-N-glycosylation and trypsin digestions are accelerated by microwave irradiation and performed sequentially in a single spin filter. Both N-glycans and peptides (including de-N-glycosylated peptides) are separately collected by filtration. The condition to effectively collect complex and heterogeneous N-glycans was established on model glycoproteins, bovine ribonuclease B, bovine fetuin, and human serum IgG. With this platform, the N-glycome, Nglycoproteome and proteome of human urine and plasma were characterized. Overall, a total of 865 and 295 Nglycosites were identified from three pairs of urine and plasma samples, respectively. Many sites were defined unambiguously as partially occupied by the detection of their nonsugar-modified peptides (128 from urine and 61 from plasma), demonstrating that partial occupancy of N-glycosylation occurs frequently. Given the likely high prevalence and variability of partial occupancy, glycoprotein quantification based exclusively on deglycosylated peptides may lead to inaccurate quantification. Molecular

show abstract

Statistical Validation of Peptide Identifications in Large-Scale Proteomics Using the Target-Decoy Database Search Strategy and Flexible Mixture Modeling

Cited by 112 publications

References 26 publications

Transferred Subgroup False Discovery Rate for Rare Post-translational Modifications Detected by Mass Spectrometry

Transferred Subgroup False Discovery Rate for Rare Post-translational Modifications Detected by Mass Spectrometry

Modes of Inference for Evaluating the Confidence of Peptide Identifications

The GlycoFilter: A Simple and Comprehensive Sample Preparation Platform for Proteomics, N-Glycomics and Glycosylation Site Assignment

Contact Info

Product

Resources

About