Statistical control of peptide and protein error rates in large-scale targeted data-independent acquisition analyses

Abstract

Liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) is the main method for high-throughput identification and quantification of peptides and inferred proteins. Within this field, data-independent acquisition (DIA) combined with peptide-centric scoring, as exemplified by the technique SWATH-MS, has emerged as a scalable method to achieve deep and consistent proteome coverage across large-scale data sets. We demonstrate that statistical concepts developed for discovery proteomics based on spectrum-centric scoring can be adapted to large-scale DIA experiments that have been analyzed with peptide-centric scoring strategies, and we provide guidance on their application. We show that optimal tradeoffs between sensitivity and specificity require careful considerations of the relationship between proteins in the samples and proteins represented in the spectral library. We propose the application of a global analyte constraint to prevent the accumulation of false positives across large-scale data sets. Furthermore, to increase the quality and reproducibility of published proteomic results, well-established confidence criteria should be reported for the detected peptide queries, peptides and inferred proteins.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Figure 1: Estimation of q-values at the peptide-query level, the peptide level and the protein level.
Figure 2: Schematic illustration of the different context-dependent error-rate estimation strategies.
Figure 3: Analyte accumulation across multiple runs.
Figure 4: Comparison between peptide queries with varying target prevalence.

References

  1. 1

    Domon, B. & Aebersold, R. Options and considerations when selecting a quantitative proteomics strategy. Nat. Biotechnol. 28, 710–721 (2010).

    CAS  PubMed  PubMed Central  Google Scholar 

  2. 2

    Chapman, J.D., Goodlett, D.R. & Masselon, C.D. Multiplexed and data-independent tandem mass spectrometry for global proteome profiling. Mass Spectrom. Rev. 33, 452–470 (2014).

    CAS  PubMed  Google Scholar 

  3. 3

    Gillet, L.C., Leitner, A. & Aebersold, R. Mass spectrometry applied to bottom-up proteomics: entering the high-throughput era for hypothesis testing. Annu. Rev. Anal. Chem. (Palo Alto Calif.) 9, 449–472 (2016).

    Google Scholar 

  4. 4

    Ting, Y.S. et al. Peptide-centric proteome analysis: an alternative strategy for the analysis of tandem mass spectrometry data. Mol. Cell. Proteomics 14, 2301–2307 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  5. 5

    Silva, J.C. et al. Quantitative proteomic analysis by accurate mass-retention-time pairs. Anal. Chem. 77, 2187–2200 (2005).

    CAS  PubMed  Google Scholar 

  6. 6

    Tsou, C.-C. et al. DIA-Umpire: comprehensive computational framework for data-independent acquisition proteomics. Nat. Methods 12, 258–264 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  7. 7

    Wang, J. et al. MSPLIT-DIA: sensitive peptide identification for data-independent acquisition. Nat. Methods 12, 1106–1108 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  8. 8

    Li, Y. et al. Group-DIA: analyzing multiple data-independent acquisition mass spectrometry data files. Nat. Methods 12, 1105–1106 (2015).

    CAS  PubMed  Google Scholar 

  9. 9

    Gillet, L.C. et al. Targeted data extraction of the MS/MS spectra generated by data-independent acquisition: a new concept for consistent and accurate proteome analysis. Mol. Cell. Proteomics 11, O111.016717 (2012).

    PubMed  PubMed Central  Google Scholar 

  10. 10

    Röst, H.L. et al. OpenSWATH enables automated, targeted analysis of data-independent acquisition MS data. Nat. Biotechnol. 32, 219–223 (2014).

    PubMed  Google Scholar 

  11. 11

    Teleman, J. et al. DIANA—algorithmic improvements for analysis of data-independent acquisition MS data. Bioinformatics 31, 555–562 (2015).

    CAS  PubMed  Google Scholar 

  12. 12

    MacLean, B. et al. Skyline: an open-source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics 26, 966–968 (2010).

    CAS  PubMed  PubMed Central  Google Scholar 

  13. 13

    Bruderer, R. et al. Extending the limits of quantitative proteome profiling with data-independent acquisition and application to acetaminophen-treated three-dimensional liver microtissues. Mol. Cell. Proteomics 14, 1400–1410 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  14. 14

    Carr, S.A. et al. Targeted peptide measurements in biology and medicine: best practices for mass-spectrometry-based assay development using a fit-for-purpose approach. Mol. Cell. Proteomics 13, 907–917 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

  15. 15

    Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate—a practical and powerful approach to multiple testing. J. R. Stat. Soc. B Stat. Methodol. 57, 289–300 (1995).

    Google Scholar 

  16. 16

    Keller, A., Nesvizhskii, A.I., Kolker, E. & Aebersold, R. Empirical statistical model to estimate the accuracy of peptide identifications made by MS/MS and database search. Anal. Chem. 74, 5383–5392 (2002).

    CAS  PubMed  Google Scholar 

  17. 17

    Choi, H. & Nesvizhskii, A.I. Semi-supervised model-based validation of peptide identifications in mass-spectrometry-based proteomics. J. Proteome Res. 7, 254–265 (2008).

    CAS  PubMed  Google Scholar 

  18. 18

    Käll, L., Storey, J.D., MacCoss, M.J. & Noble, W.S. Posterior error probabilities and false discovery rates: two sides of the same coin. J. Proteome Res. 7, 40–44 (2008).

    PubMed  Google Scholar 

  19. 19

    Genovese, C. & Wasserman, L. Operating characteristics and extensions of the false discovery rate procedure. J. R. Stat. Soc. B Stat. Methodol. 64, 499–517 (2002).

    Google Scholar 

  20. 20

    Iyer, V. & Sarkar, S. An adaptive single-step FDR procedure with applications to DNA microarray analysis. Biom. J. 49, 127–135 (2007).

    PubMed  Google Scholar 

  21. 21

    Storey, J.D. The positive false discovery rate: a Bayesian interpretation and the q-value. Ann. Stat. 31, 2013–2035 (2003).

    Google Scholar 

  22. 22

    Nesvizhskii, A.I. A survey of computational methods and error-rate estimation procedures for peptide and protein identification in shotgun proteomics. J. Proteomics 73, 2092–2123 (2010).

    CAS  PubMed  PubMed Central  Google Scholar 

  23. 23

    Käll, L., Canterbury, J.D., Weston, J., Noble, W.S. & MacCoss, M.J. Semi-supervised learning for peptide identification from shotgun proteomics data sets. Nat. Methods 4, 923–925 (2007).

    PubMed  Google Scholar 

  24. 24

    Serang, O. & Noble, W. A review of statistical methods for protein identification using tandem mass spectrometry. Stat. Interface 5, 3–20 (2012).

    PubMed  PubMed Central  Google Scholar 

  25. 25

    The, M., Tasnim, A. & Käll, L. How to talk about protein-level false discovery rates in shotgun proteomics. Proteomics 16, 2461–2469 (2016).

    CAS  PubMed  PubMed Central  Google Scholar 

  26. 26

    Shteynberg, D. et al. iProphet: multilevel integrative analysis of shotgun proteomic data improves peptide and protein identification rates and error estimates. Mol. Cell. Proteomics 10, M111.007690 (2011).

    PubMed  PubMed Central  Google Scholar 

  27. 27

    Reiter, L. et al. Protein identification false discovery rates for very large proteomics data sets generated by tandem mass spectrometry. Mol. Cell. Proteomics 8, 2405–2417 (2009).

    CAS  PubMed  PubMed Central  Google Scholar 

  28. 28

    Savitski, M.M., Wilhelm, M., Hahne, H., Kuster, B. & Bantscheff, M. A scalable approach for protein false discovery rate estimation in large proteomic data sets. Mol. Cell. Proteomics 14, 2394–2404 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  29. 29

    The, M., MacCoss, M.J., Noble, W.S. & Käll, L. Fast and accurate protein false discovery rates on large-scale proteomics data sets with Percolator 3.0. J. Am. Soc. Mass Spectrom. 27, 1719–1727 (2016).

    CAS  PubMed  PubMed Central  Google Scholar 

  30. 30

    Choi, H., Ghosh, D. & Nesvizhskii, A.I. Statistical validation of peptide identifications in large-scale proteomics using the target-decoy database search strategy and flexible mixture modeling. J. Proteome Res. 7, 286–292 (2008).

    CAS  PubMed  Google Scholar 

  31. 31

    Ahrens, C.H., Brunner, E., Qeli, E., Basler, K. & Aebersold, R. Generating and navigating proteome maps using mass spectrometry. Nat. Rev. Mol. Cell Biol. 11, 789–801 (2010).

    CAS  PubMed  Google Scholar 

  32. 32

    Reiter, L. et al. mProphet: automated data processing and statistical validation for large-scale SRM experiments. Nat. Methods 8, 430–435 (2011).

    CAS  PubMed  Google Scholar 

  33. 33

    Karlsson, C., Malmström, L., Aebersold, R. & Malmström, J. Proteome-wide selected reaction monitoring assays for the human pathogen Streptococcus pyogenes. Nat. Commun. 3, 1301 (2012).

    PubMed  PubMed Central  Google Scholar 

  34. 34

    Schubert, O.T. et al. The Mtb proteome library: a resource of assays to quantify the complete proteome of Mycobacterium tuberculosis. Cell Host Microbe 13, 602–612 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  35. 35

    Picotti, P. et al. A complete mass spectrometric map of the yeast proteome applied to quantitative trait analysis. Nature 494, 266–270 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  36. 36

    Rosenberger, G. et al. A repository of assays to quantify 10,000 human proteins by SWATH-MS. Sci. Data 1, 140031 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

  37. 37

    Collins, B.C. et al. Multi-laboratory assessment of reproducibility, qualitative and quantitative performance of SWATH–mass spectrometry. Nat. Commun. 8, DOI: 10.1038/s41467-017-00249-5 (2017).

  38. 38

    Liu, Y. et al. Quantitative variability of 342 plasma proteins in a human twin population. Mol. Syst. Biol. 11, 786 (2015).

    PubMed  PubMed Central  Google Scholar 

  39. 39

    Selevsek, N. et al. Reproducible and consistent quantification of the Saccharomyces cerevisiae proteome by SWATH-MS. Mol. Cell. Proteomics 14, 739–749 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  40. 40

    Guo, T. et al. Rapid mass spectrometric conversion of tissue biopsy samples into permanent quantitative digital proteome maps. Nat. Med. 21, 407–413 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  41. 41

    Schubert, O.T. et al. Absolute proteome composition and dynamics during dormancy and resuscitation of Mycobacterium tuberculosis. Cell Host Microbe 18, 96–108 (2015).

    CAS  PubMed  Google Scholar 

  42. 42

    Schubert, O.T. et al. Building high-quality assay libraries for targeted analysis of SWATH-MS data. Nat. Protoc. 10, 426–441 (2015).

    CAS  PubMed  Google Scholar 

  43. 43

    Storey, J.D. & Tibshirani, R. Statistical significance for genome-wide studies. Proc. Natl. Acad. Sci. USA 100, 9440–9445 (2003).

    CAS  PubMed  Google Scholar 

  44. 44

    Serang, O. & Käll, L. Solution to statistical challenges in proteomics is more statistics, not less. J. Proteome Res. 14, 4099–4103 (2015).

    CAS  PubMed  Google Scholar 

  45. 45

    Blattmann, P., Heusel, M. & Aebersold, R. SWATH2stats: an R/Bioconductor package to process and convert quantitative SWATH-MS proteomics data for downstream analysis tools. PLoS One 11, e0153160 (2016).

    PubMed  PubMed Central  Google Scholar 

  46. 46

    Tsou, C.-C., Tsai, C.F., Teo, G.C., Chen, Y.J. & Nesvizhskii, A.I. Untargeted, spectral library-free analysis of data-independent acquisition proteomics data generated using Orbitrap mass spectrometers. Proteomics 16, 2257–2271 (2016).

    CAS  PubMed  PubMed Central  Google Scholar 

  47. 47

    Keller, A., Bader, S.L., Shteynberg, D., Hood, L. & Moritz, R.L. Automated validation of results and removal of fragment ion interferences in targeted analysis of data-independent acquisition mass spectrometry (MS) using SWATHProphet. Mol. Cell. Proteomics 14, 1411–1418 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  48. 48

    Gupta, N. & Pevzner, P.A. False discovery rates of protein identifications: a strike against the two-peptide rule. J. Proteome Res. 8, 4173–4181 (2009).

    CAS  PubMed  PubMed Central  Google Scholar 

  49. 49

    Muntel, J. et al. Advancing urinary protein biomarker discovery by data-independent acquisition on a quadrupole-orbitrap mass spectrometer. J. Proteome Res. 14, 4752–4762 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  50. 50

    Vizcaíno, J.A. et al. The PRoteomics IDEntifications (PRIDE) database and associated tools: status in 2013. Nucleic Acids Res. 41, D1063–D1069 (2013).

    PubMed  Google Scholar 

Download references

Acknowledgements

Please note that M.H., C.L.H., Y.L., M.J.M., B.X.M., A.I.N., P.G.A.P., L.R., H.L.R., S.T. and Y.S.T. were added to the author list in alphabetical order. We thank the authors of the SWATH-MS interlaboratory study and of the human blood plasma data set for providing the data to conduct this study. We also thank the Scientific IT Support (ID SIS) and the high-performance computing (HPC) teams of ETH Zurich for support and maintenance of the computing infrastructure. M.H. was supported by a grant from the Institut Mérieux; A.I.N. was funded by the US National Institutes of Health (NIH; grant R01GM094231); H.L.R. was funded by the Swiss National Science Foundation (SNSF; grant P2EZP3 162268); B.C.C. was supported by a SNSF Ambizione grant (PZ00P3_161435); and R.A. was supported by ERC Proteomics v3.0 (AdG-233226 Proteomics v.3.0) and AdG-670821 Proteomics 4D), the PhosphonetX project of SystemsX.ch and the Swiss National Science Foundation (SNSF) grant 31003A_166435.

Author information

Affiliations

Authors

Contributions

G.R., I.B. and R.A. wrote the paper with feedback from all authors; G.R. and B.C.C. developed the methods; I.B. analyzed the data set; U.S. and G.R. implemented the PyProphet extension; M.H., C.L.H., Y.L., M.J.M., B.X.M., A.I.N., P.G.A.P., L.R., H.L.R., S.T. and Y.S.T. provided critical input on the project; and B.C.C. and R.A. designed and supervised the study.

Corresponding authors

Correspondence to Ben C Collins or Ruedi Aebersold.

Ethics declarations

Competing interests

C.L.H. and S.T. are employees of SCIEX, which operates in the field of quantitative proteomics by data-independent acquisition covered by the article. M.J.M. is a paid consultant for Thermo Fisher Scientific, which operates in the field of quantitative proteomics by data-independent acquisition covered by the article. L.R. is employee of Biognosys AG, which operates in the field of quantitative proteomics by data-independent acquisition covered by the article. R.A. holds shares of Biognosys AG.

Integrated supplementary information

Supplementary Figure 1 Estimation of q-values at the peptide-query level, the peptide level and the protein level.

The peptide-query-level (left), peptide-level (middle) and protein-level (right) discriminant score density plots (a) and p-value histograms (b) for one DIA run of the SWATH-MS interlaboratory study that was analyzed with the combined human assay library (CAL) are shown. a) The distributions indicate a large (false target)/(total target) ratio (π0 0.6) on the peptide-query level. The q-value estimation was adapted for peptide and protein level by using the best scoring peak group per peptide or protein across all samples for both targets and decoys. The (false target)/(total target) ratio decreases slightly on peptide level and more on protein level (π0 0.5), compared to the peptide-query level. b) On the peptide-query level and the peptide level, the estimation of π0 is anticonservative, indicated by a lower density of p-values after the p-value threshold of λ=0.4. On the protein level, the estimation of π0 is more accurate with a consistent density of p-values.

Supplementary Figure 2 Influence of protein length on the peptide-query-level and protein-level q-value estimation.

a) Protein length distribution of all proteins in the combined human assay library (CAL), all proteins inferred at 1% peptide-query-level FDR in the global context of all 229 DIA runs of the SWATH-MS interlaboratory comparison study, and all proteins inferred at 1% global protein FDR respectively. b) Histogram of protein length distribution for the differently filtered protein subsets of the CAL. The distributions show that there is no bias for protein length when selecting the best peak group as proxy for protein-level q-value estimation.

Supplementary Figure 3 Decoy accumulation across multiple runs.

The number of cumulatively detected peak group decoys (a), peptide decoys (b) and protein decoys (c) is shown for 229 DIA runs of the SWATH-MS interlaboratory study data set.

Supplementary Figure 4 Analyte accumulation across multiple runs (5% FDR).

The number of cumulatively detected peak groups (a), peptides (b) and proteins (c) is shown for 229 DIA runs of the SWATH-MS interlaboratory study data set.

Supplementary Figure 5 Decoy accumulation across multiple runs (5% FDR).

The number of cumulatively detected peak group decoys (a), peptide decoys (b) and protein decoys (c) is shown for 229 DIA runs of the SWATH-MS interlaboratory study data set.

Supplementary Figure 6 Combined human and M. tuberculosis spectral library analysis.

a) The peptide-level discriminant score density of human targets, human decoys, M. tuberculosis (Mtb) targets, and Mtb decoys is shown for global analysis of the 229 DIA runs of the SWATH-MS interlaboratory study data set applying the combined human and Mtb spectral library. The Mtb targets and decoys show a similar distribution compared to the human decoys and the fraction of false human targets. The number of cumulatively detected peptides is shown for human targets (b), human decoys (c), Mtb targets (d), and Mtb decoys (e) from the combined human and Mtb spectral library with different error rate control strategies. The Mtb decoy to target ratio is 0.82, explaining the absolute higher number of the accumulated Mtb targets.

Supplementary Figure 7 Analyte accumulation across multiple runs in the plasma data set (1% FDR).

The number of cumulatively detected peak groups (a), peptides (b) and proteins (c) is shown for the 246 DIA runs of the plasma data set analyzed with the nonparametric model for q-value estimation.

Supplementary information

Supplementary Text and Figures

Supplementary Figures 1–7, Supplementary Table 1 and Supplementary Notes 1–6. (PDF 1409 kb)

Life Sciences Reporting Summary (PDF 71 kb)

Source data

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Rosenberger, G., Bludau, I., Schmitt, U. et al. Statistical control of peptide and protein error rates in large-scale targeted data-independent acquisition analyses. Nat Methods 14, 921–927 (2017). https://doi.org/10.1038/nmeth.4398

Download citation

Further reading

Search

Quick links

Sign up for the Nature Briefing newsletter for a daily update on COVID-19 science.
Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing