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:

Correcting for cell-type heterogeneity in DNA methylation: a comprehensive evaluation

Nature Methods volume 14, pages 218–219 (2017)Cite this article

Subjects

This article has been updated

Rahmani et al. reply:

Zheng et al.1 discuss potential pitfalls in our evaluation of ReFACTor2, a reference-free method to account for cell-type heterogeneity. Below, we reproduce their analysis and demonstrate that conclusions cannot be drawn on the basis of their results owing to conceptual and technical flaws in their analysis. We show with our reanalysis and further evidence from experiments on a total of 10 data sets that ReFACTor has improved performance over alternative methods, including the reference-based method of Houseman et al.3.

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

Change history

  • 14 March 2017

    In the version of this article initially published, some numbers in Table 1 did not appear in boldface. In the HTML originally posted online, the author affiliation for Elior Rahmani was incorrect; Rahmani is affiliated with only the Tel-Aviv University, Israel. The Supplementary Information file has been replaced to correct for additional callouts of Supplementary Notes in the Supplementary Figure legends. The errors have been corrected in the HTML and PDF files as of 14 March 2017.

References

  1. Zheng, S.C. et al. Nat. Methods 14, 216–217 (2017).

    Article  CAS  Google Scholar 

  2. Rahmani, E. et al. Nat. Methods 13, 443–445 (2016).

    Article  CAS  Google Scholar 

  3. Houseman, E.A. et al. BMC Bioinformatics 13, 86 (2012).

    Article  Google Scholar 

  4. Leek, J.T. & Storey, J.D. PLoS Genet. 3, e161 (2007).

    Article  Google Scholar 

  5. Teschendorff, A.E. et al. Nat. Commun. 7, 10478 (2016).

    Article  CAS  Google Scholar 

  6. Koestler, D.C. et al. BMC Bioinformatics 17, 120 (2016).

    Article  Google Scholar 

  7. Teschendorff, A.E., Zhuang, J. & Widschwendter, M. Bioinformatics 27, 1496–1505 (2011).

    Article  CAS  Google Scholar 

  8. Tan, Q. et al. BMC Genomics 15, 1062 (2014).

    Article  Google Scholar 

  9. Liu, Y. et al. Nat. Biotechnol. 31, 142–147 (2013).

    Article  CAS  Google Scholar 

  10. Hannon, E. et al. Genome Biol. 17, 176 (2016).

    Article  Google Scholar 

Download references

Acknowledgements

This research was partially supported by the Edmond J. Safra Center for Bioinformatics at Tel-Aviv University, the Israel Science Foundation (1425/13 to E.R. and E.H.), US National Science Foundation grant 1331176 and United States Israel Binational Science Foundation grant 2012304 (to E.R., Y.B. and E.H.). E.R. was supported by Len Blavatnik and the Blavatnik Research Foundation. N.Z. was supported in part by a US National Institutes of Health (NIH) career development award from the NHLBI (K25HL121295). C.E., S.H., D.H., J.G., S.O. and E.G.B. were supported by the Sandler Family Foundation, the American Asthma Foundation, Hind Distinguished Professorships and NIH grants 1P60MD006902, 1R01HL117004, R21ES24844, R01Hl128439 and TRDRP 24RT-0025. E.E. was supported by NSF grants 1065276, 1302448, 1320589 and 1331176 and NIH grants R01-GM083198, R01-ES021801, R01-MH101782, R01-ES022282 and U54EB020403.

Author information

Author notes

  1. Joshua Galanter

    Present address: Present address: Genentech, South San Francisco, California, USA.,

Authors and Affiliations

  1. Blavatnik School of Computer Science, Tel-Aviv University, Tel Aviv, Israel

    Elior Rahmani

  2. Department of Medicine, University of California, San Francisco, San Francisco, California, USA

    Noah Zaitlen, Celeste Eng, Donglei Hu, Joshua Galanter, Sam Oh & Esteban G Burchard

  3. Department of Computer Science and Applied Mathematics, The Weizmann Institute of Science, Rehovot, Israel

    Yael Baran

  4. Department of Bioengineering and Therapeutic Science, University of California, San Francisco, San Francisco, California, USA

    Joshua Galanter & Esteban G Burchard

  5. Department of Computer Science, University of California, Los Angeles, Los Angeles, California, USA

    Eleazar Eskin & Eran Halperin

  6. Department of Human Genetics, University of California, Los Angeles, Los Angeles, California, USA

    Eleazar Eskin

  7. Department of Biomedical Data Science, Stanford University, Palo Alto, California, USA

    James Zou

  8. Department of Anesthesiology and Perioperative Medicine, University of California, Los Angeles, Los Angeles, California, USA

    Eran Halperin

Authors
  1. Elior Rahmani
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Noah Zaitlen
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Yael Baran
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Celeste Eng
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Donglei Hu
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Joshua Galanter
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Sam Oh
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Esteban G Burchard
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Eleazar Eskin
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. James Zou
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Eran Halperin
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Eran Halperin.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Integrated supplementary information

Supplementary Figure 1 Evaluation of the number of expected spurious associations when using a small number of controls in EWAS.

The histogram of significant associations found across 100 EWAS experiments on the data used by Zheng et al. for constructing the gold standard list of “true positives”. The red line marks 23,258, the number of sites defined in the “true positives” list by Zheng et al.

Supplementary Figure 2 Capturing cell-type composition in breast cancer data using ReFACTor.

(a) A reconstruction of Figure S1 from Zheng et al., showing correlation of the cell-types and disease status (N/C) with each of the first 25 principal components of the data (n=355). Here, as well as in the following subfigures, the colors correspond to the logarithm of the P-values of the correlations. (b) The correlation of the first 25 ReFACTor components with the cell-types and disease status, as well as with the variation of the disease status that is independent of the cell composition (Adj. N/C). (c) The correlation of the first 25 SVA components (SVs) with the cell-types and with the unadjusted and adjusted disease status. (d) The mean R2 levels, across the nine estimated cell-types, of linear models fitted for each cell-type using an increasing number of ReFACTor component and using an increasing number of SVs. For any given number of components, ReFACTor has better R2 level than SVA.

Supplementary Figure 3 Capturing cell-type composition variation in the GALA II dataset.

(a)-(d) R2 values of the linear model predicting flow-cytometric estimates for blood cell-types, as a function of the number of ReFACTor components included in the model (blue data points and lines) for the GALA II dataset (n=84). Horizontal blue lines indicate the R2 values of the model using the ReFACTor components with significant likelihood ratio test (LRT) P-values (significant components are marked with squares). The reference-based estimates of the entire GALA II dataset (n=560) were used to determine the number of significant ReFACTor components. Horizontal orange lines indicate the performance of the reference-based method. (e) The mean R2 level over all cell-types.

Supplementary Figure 4 Capturing cell-type composition variation in the Koestler et al. dataset.

(a)-(f) R2 values of the linear model predicting flow-cytometric estimates for blood cell-types, as a function of the number of ReFACTor components included in the model (blue data points and lines) for the Koestler et al. dataset (n=18). Horizontal blue lines indicate the R2 values of the model using the ReFACTor components with significant likelihood ratio test (LRT) P-values (significant components are marked with squares). Horizontal orange lines indicate the performance of the reference-based method. (g) The mean R2 level over all cell-types.

Supplementary Figure 5 Performance for capturing cell-type composition in small data is highly variable.

(a) Sampling 100 subsets of 18 individuals with cell counts from the GALA II dataset (n=84) reveals that the performance (measured in mean R2 across all cell-types) of both ReFACTor and the reference-based method are highly variable due to the small number of samples used. (b) Distribution of the performance after sampling 100 subsets of 15 individuals from the Koestler et al. data (n=18).

Supplementary Figure 6 RMT estimates of the dimension in data as a function of the sample size.

The estimated dimensions by the RMT method (Teschendorff et al. 2011) as a function of the number of samples (each time adding a new randomly selected sample) using (a) the GALA II dataset (n=560), (b) a dataset by Liu et al. (n=686) and (c)-(d) two independent datasets by Hannon et al (n=675 and n=847). Linear regression lines (indicated in red) demonstrate a nearly perfect linear relation (P-value<10−93 in all plots).

Supplementary information

Supplementary Text and Figures

Supplementary Figures 1–6, Supplementary Tables 1–6, Supplementary Methods and Supplementary Notes 1–5 (PDF 720 kb)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Rahmani, E., Zaitlen, N., Baran, Y. et al. Correcting for cell-type heterogeneity in DNA methylation: a comprehensive evaluation. Nat Methods 14, 218–219 (2017). https://doi.org/10.1038/nmeth.4190

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nmeth.4190

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