Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Correspondence
  • Published:

Avoiding abundance bias in the functional annotation of posttranslationally modified proteins

Nature Methods volume 12, pages 1003–1004 (2015)Cite this article

Subjects

To the Editor

Identification of post-translational modifications (PTMs) by mass spectrometry is biased toward abundant proteins, skewing our understanding of PTMs in favor of readily detected proteins. We developed a method and web tool (https://agotool.sund.ku.dk) to account for this protein-abundance bias in Gene Ontology (GO)-term enrichment analyses of PTM data sets.

This is a preview of subscription content, access via your institution

Relevant articles

Open Access articles citing this article.

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

Prices may be subject to local taxes which are calculated during checkout

Figure 1: Improved functional annotation of protein PTMs.

References

  1. Ashburner, M. et al. Nat. Genet. 25, 25–29 (2000).

    Article  CAS  Google Scholar 

  2. Wagner, G.R. & Payne, R.M. J. Biol. Chem. 288, 29036–29045 (2013).

    Article  CAS  Google Scholar 

  3. Weinert, B.T. et al. Cell Rep. 4, 842–851 (2013).

    Article  CAS  Google Scholar 

  4. Castaño-Cerezo, S. et al. Mol. Syst. Biol. 10, 762 (2014).

    Article  Google Scholar 

  5. Weinert, B.T. et al. Mol. Cell 51, 265–272 (2013).

    Article  CAS  Google Scholar 

  6. Paik, W.K., Pearson, D., Lee, H.W. & Kim, S. Biochim. Biophys. Acta 213, 513–522 (1970).

    Article  CAS  Google Scholar 

  7. Weinert, B.T. et al. Mol. Syst. Biol. 10, 716 (2014).

    Article  Google Scholar 

  8. Choudhary, C. et al. Science 325, 834–840 (2009).

    Article  CAS  Google Scholar 

  9. Henriksen, P. et al. Mol. Cell. Proteomics 11, 1510–1522 (2012).

    Article  Google Scholar 

  10. Rardin, M.J. et al. Cell Metab. 18, 920–933 (2013).

    Article  CAS  Google Scholar 

  11. Weinert, B.T. et al. Sci. Signal. 4, ra48 (2011).

    Article  CAS  Google Scholar 

  12. Zhao, S. et al. Science 327, 1000–1004 (2010).

    Article  CAS  Google Scholar 

  13. Schwanhäusser, B. et al. Nature 473, 337–342 (2011).

    Article  Google Scholar 

Download references

Acknowledgements

We thank the members of the Department of Proteomics and the Department of Disease Systems Biology at the Center for Protein Research for helpful discussions. C.C. is supported by the Hallas Møller Investigator Fellowship from the Novo Nordisk Foundation. The Center for Protein Research is funded by a generous grant from the Novo Nordisk Foundation (NNF14CC0001).

Author information

Author notes

  1. Christian Schölz and David Lyon: These authors contributed equally to this work.

Authors and Affiliations

  1. Department of Proteomics, The Novo Nordisk Foundation Center for Protein Research, University of Copenhagen, Copenhagen, Denmark

    Christian Schölz, Chunaram Choudhary & Brian T Weinert

  2. Department of Disease Systems Biology, The Novo Nordisk Foundation Center for Protein Research, University of Copenhagen, Copenhagen, Denmark

    David Lyon, Jan C Refsgaard & Lars Juhl Jensen

Authors
  1. Christian Schölz
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. David Lyon
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Jan C Refsgaard
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Lars Juhl Jensen
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Chunaram Choudhary
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Brian T Weinert
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding authors

Correspondence to Chunaram Choudhary or Brian T Weinert.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Integrated supplementary information

Supplementary Figure 1 Abundance distributions of post-translationally modified proteins.

The histograms show the distributions of protein abundances as determined using an intensity-based absolute quantification (iBAQ) method1. Experimentally detected proteins (proteins) and posttranslationally modified proteins (modified proteins) are shown. The distributions of the indicated posttranslationally modified proteins were compared to the experimentally detected proteins by Wilcoxon test (P-value). The number of analyzed proteins is indicated in parenthesis.

1 Schwanhausser, B. et al. Global quantification of mammalian gene expression control. Nature 473, 337–342 (2011).

Supplementary Figure 2 Overlap between different GO-term enrichment analyses of post-translationally modified proteins.

The Venn diagrams show the overlap between significantly (P < 0.01) enriched GO terms and UniProt keywords identified by comparing the indicated groups of modified proteins to the genome, observed proteome, and abundance-corrected (corrected) proteome. Note that some terms were found using the corrected proteome that were not found using the genome and/or observed proteome.

Supplementary Figure 3 The increased specificity of the abundance-corrected analysis is partly due to the smaller sample size of the abundance-corrected proteome compared with that of the observed proteome.

(a) The model illustrates the method used to account for differences in sample size. The number of term-associating and total proteins in the abundance-corrected (corrected) proteome were increased proportionally to match those in the observed proteome (observed), which we refer to as the adjusted-corrected proteome (the frequency of term association is identical in the corrected and adjusted-corrected proteomes, but the sample size is artificially increased in the adjusted-corrected proteome to mimic the statistical power of the observed proteome. This method was only applied here and is not a feature of the web tool because terms identified using the adjusted-corrected proteome are not necessarily significant since the sample size was artificially increased). This allowed us to determine whether terms identified using the observed proteome were absent when using the corrected proteome because the number of observations decreased, thereby reducing the statistical significance. We defined the total increase in specificity as the number of terms that were found using the observed proteome, but not the corrected proteome. The number of these terms that became significant again by using the adjusted-corrected proteome was attributed to a loss of significance due to the sample size. Thus, the fraction of increased specificity due to reduced sample size is equal to the number of terms regained by using the adjusted-corrected proteome divided by the total increase in specificity. (b) The Venn diagrams show the overlap between significantly enriched terms (P<0.01) using the observed, corrected, or adjusted-corrected proteomes to compare GO term associations with the indicated groups of modified proteins. The fraction of increased specificity attributed to sample size is show together with the numbers used to calculate this value. Note that increasing the sample size did not enable the detection of significantly enriched GO terms or KWs for succinylation in yeast.

Supplementary Figure 4 KWs describing central metabolism were removed because of abundance bias and not sample size.

Based on the analysis shown in Supplementary Fig. 3 we determined that decreased keyword association in the abundance-corrected (corrected) analysis could either be attributed to sample size (keyword is significantly enriched after artificially increasing the sample size in the adjusted-corrected proteome) or to abundance-bias (increasing the sample size has no effect on the significance of keyword enrichment). Here we show significantly (P < 0.01) enriched UniProt keywords for acetylated proteins which were removed in the corrected analysis due to abundance-bias or sample size. The Venn diagrams show significantly enriched keywords for acetylation in yeast and HeLa cells based on comparison to the observed proteome, corrected proteome, or adjusted-corrected proteome. The fold enrichment (Enrichment) and Fisher P value (P value) based on the observed proteome analysis are shown. The keyword “Glycolysis” was one of the most highly enriched keywords in the observed proteome which was attributed to abundance-bias.

Supplementary information

Supplementary Text and Figures

Supplementary Figures 1–4 and Supplementary Methods (PDF 2645 kb)

Supplementary Table 1

PTM mapping in yeast cells (XLSX 5975 kb)

Supplementary Table 2

PTM mapping in HeLa cells (XLSX 7602 kb)

Supplementary Data

GO term and UniProt Keyword enrichment analysis (ZIP 9697 kb)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Schölz, C., Lyon, D., Refsgaard, J. et al. Avoiding abundance bias in the functional annotation of posttranslationally modified proteins. Nat Methods 12, 1003–1004 (2015). https://doi.org/10.1038/nmeth.3621

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nmeth.3621

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing