Methodology

Increased correlation between methylation sites in epigenome-wide replication studies: impact on analysis and results

Maja Popovic

*Author for correspondence: Tel.: +39 (0) 116334628;

E-mail Address: maja_popovic@hotmail.com

Department of Medical Sciences, University of Turin & CPO Piemonte, Turin, Italy

Search for more papers by this author

Francesca Fasanelli

Department of Medical Sciences, University of Turin & CPO Piemonte, Turin, Italy

Search for more papers by this author

Valentina Fiano

Department of Medical Sciences, University of Turin & CPO Piemonte, Turin, Italy

Search for more papers by this author

Annibale Biggeri

Department of Statistics, Computer Science, Applications «G. Parenti», University of Florence, Florence, Italy

Search for more papers by this author

Lorenzo Richiardi

Department of Medical Sciences, University of Turin & CPO Piemonte, Turin, Italy

Search for more papers by this author

Published Online:6 Nov 2017https://doi.org/10.2217/epi-2017-0073

View article

Abstract

Aim: To show that an increased correlation between CpGs after selection through an epigenome-wide association studies (EWAS) might translate into biased replication results. Methods: Pairwise correlation coefficients between CpGs selected in two published EWAS, the top hits replication, Bonferroni p-values, Benjamini–Hochberg (BH) false discovery rate (FDR) and directional FDR r-values were calculated in the NINFEA cohort data. Exposures’ random permutations were performed to show the empirical p-value distributions. Results: The average pairwise correlation coefficients between CpGs were enhanced after selection for the replication (e.g., from 0.12 at genome-wide level to 0.26 among the selected CpGs), affecting the empirical p-value distributions and the usual multiple testing control. Conclusion: Bonferroni and Benjamini–Hochberg FDR are inappropriate for the EWAS replication phase, and methods that account for the underlying correlation need to be used.

Keywords:

References

1 Eckhardt F, Lewin J, Cortese R et al. DNA methylation profiling of human chromosomes 6, 20 and 22. Nat. Genet. 38(12), 1378–1385 (1987).Crossref, Google Scholar
2 Gardiner-Garden M, Frommer M. CpG islands in vertebrate genomes. J. Mol. Biol. 196(2), 261–282 (1987).Crossref, Medline, CAS, Google Scholar
3 Ong ML, Holbrook JD. Novel region discovery method for Infinium 450 K DNA methylation data reveals changes associated with aging in muscle and neuronal pathways. Aging Cell 13(1), 142–155 (2014).Crossref, Medline, CAS, Google Scholar
4 Jaffe AE, Murakami P, Lee H et al. Bump hunting to identify differentially methylated regions in epigenetic epidemiology studies. Int. J. Epidemiol. 41(1), 200–209 (2012).Crossref, Medline, Google Scholar
5 Sofer T, Schifano ED, Hoppin JA, Hou L, Baccarelli AA. A-clustering: a novel method for the detection of co-regulated methylation regions, and regions associated with exposure. Bioinformatics 29(22), 2884–2891 (2013).Crossref, Medline, CAS, Google Scholar
6 Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics 9, 559 (2008).Crossref, Medline, Google Scholar
7 Lin X, Barton S, Holbrook JD. How to make DNA methylome wide association studies more powerful. Epigenomics 8(8), 1117–1129 (2016).Link, CAS, Google Scholar
8 Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B. 57(1), 289–300 (1995).Google Scholar
9 Heller R, Bogomolov M, Benjamini Y. Deciding whether follow-up studies have replicated findings in a preliminary large-scale omics study. Proc. Natl Acad. Sci. USA 111(46), 16262–16267 (2014).Crossref, Medline, CAS, Google Scholar
10 Sofer T, Heller R, Bogomolov M et al. A powerful statistical framework for generalization testing in GWAS, with application to the HCHS/SOL. Genet. Epidemiol. 41, 251–258 (2017).Crossref, Medline, Google Scholar
11 Joubert BR, Felix JF, Yousefi P et al. DNA Methylation in newborns and maternal smoking in pregnancy: genome-wide consortium meta-analysis. Am. J. Hum. Genet. 98(4), 680–696 (2016).Crossref, Medline, CAS, Google Scholar
12 Yousefi P, Huen K, Davé V, Barcellos L, Eskenazi B, Holland N. Sex differences in DNA methylation assessed by 450 K BeadChip in newborns. BMC Genomics 16, 911 (2015).Crossref, Medline, Google Scholar
13 Rahmani E, Zaitlen N, Baran Y et al. Sparse PCA corrects for cell type heterogeneity in epigenome-wide association studies. Nat. Methods 13(5), 443–445 (2016).Crossref, Medline, CAS, Google Scholar
14 Richiardi L, Baussano I, Vizzini L et al. Feasibility of recruiting a birth cohort through the Internet: the experience of the NINFEA cohort. Eur. J. Epidemiol. 22, 831–837 (2007).Crossref, Medline, Google Scholar
15 Du P, Zhang X, Huang C-C et al. Comparison of beta-value and M-value methods for quantifying methylation levels by microarray analysis. BMC Bioinformatics 11, 587 (2010).Crossref, Medline, CAS, Google Scholar
16 R Core Team. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria (2017). www.R-project.org/.Google Scholar
17 Fisher RA. Frequency distribution of the values of the correlation coefficient in samples of an indefinitely large population. Biometrika 10(4), 507–521 (1915).Google Scholar
18 Zeileis A. Econometric computing with HC and HAC covariance matrix estimators. J. Stat. Softw. 11(10), 1–17 (2004).Crossref, Google Scholar
19 Houseman EA, Molitor J, Marsit CJ. Reference-free cell mixture adjustments in analysis of DNA methylation data. Bioinformatics 30(10), 1431–1439 (2014).Crossref, Medline, CAS, Google Scholar
20 McGregor K, Bernatsky S, Colmegna I et al. An evaluation of methods correcting for cell-type heterogeneity in DNA methylation studies. Genome Biol. 17, 84 (2016).Crossref, Medline, Google Scholar
21 Massey FJJ. The Kolmogorov–Smirnov test for goodness of fit. J. Am. Stat. Assoc. 46, 68–78 (1951).Crossref, Google Scholar
22 Razali NM, Wah YB. Power comparisons of Shapiro–Wilk, Kolmogorov–Smirnov, Lilliefors and Anderson–Darling tests. J. Stat. Modeling Anal 2, 21–33 (2011).Google Scholar
23 Anderson TW, Darling DA. A test of goodness of fit. J. Am. Stat. Assoc. 49(268), 765–769 (1954).Crossref, Google Scholar
24 Stephens MA. EDF statistics for goodness of fit and some comparisons. J. Am. Stat. Assoc. 69, 730–737 (1974).Crossref, Google Scholar
25 Warnes GR, Bolker B, Lumley T. gtools: various R programming Tools (2015). https://CRAN.R-project.org/package=gtools.Google Scholar
26 Komsta L, Novomestky F. Moments: moments, cumulants, skewness, kurtosis and related tests (2015). https://CRAN.R-project.org/package=moments.Google Scholar
27 Revolution Analytics, Weston S. Foreach: provides foreach looping construct for R (2015). https://CRAN.R-project.org/package=foreach.Google Scholar
28 Heller R, Bogomolov M, Benjamini Y. Deciding whether follow-up studies have replicated findings in a preliminary large-scale “omics’ study”. www.runmycode.org/companion/view/542.Google Scholar
29 Lin MF, Lucas HC, Shmueli G. Too big to fail: large samples and the p-value problem. Inform. Syst. Res. 24, 906–917 (2013).Crossref, Google Scholar
30 Morales E, Vilahur N, Salas LA et al. Genome-wide DNA methylation study in human placenta identifies novel loci associated with maternal smoking during pregnancy. Int. J. Epidemiol. 45(5), 1644–1655 (2016).Crossref, Medline, Google Scholar
31 Gruzieva O, Xu CJ, Breton CV et al. Epigenome-wide meta-analysis of methylation in children related to prenatal NO₂ air pollution exposure. Environ. Health Perspect. 125(1), 104–110 (2017).Crossref, Medline, CAS, Google Scholar
32 Good P. Permutation Tests: A Practical Guide To Resampling Methods For Testing Hypotheses (2nd edition). Springer-Verlag, NY, USA (1994).Crossref, Google Scholar
33 Conneely KN, Boehnke M. So many correlated tests, so little time! Rapid adjustment of P-values for multiple correlated tests. Am. J. Hum. Genet. 81(6), 1158–1168 (2007).Crossref, Medline, CAS, Google Scholar
34 Nyholt DR. A simple correction for multiple testing for single-nucleotide polymorphisms in linkage disequilibrium with each other. Am. J. Hum. Genet. 74(4), 765–769 (2004).Crossref, Medline, CAS, Google Scholar
35 Dudbridge F, Koeleman BPC. Efficient computation of significance levels for multiple associations in large studies of correlated data, including genomewide association studies. Am. J. Hum. Genet. 75(3), 424–435 (2004).Crossref, Medline, CAS, Google Scholar
36 Yekutieli D, Benjamini Y. Resampling-based false discovery rate controlling multiple test procedures for correlated test statistics. J. Stat. Plan. Infer. 82, 171–196 (1999).Crossref, Google Scholar
37 Benjamini Y, Yekutieli D. The control of the false discovery rate in multiple testing under dependency. Ann. Stat. 29(4), 1165–1188 (2001).Crossref, Google Scholar
38 Van Iterson M, van Zwet EW, Heijmans BT, BIOS Consortium. Controlling bias and inflation in epigenome- and transcriptome-wide association studies using the empirical null distribution. Genome Biol. 18(1), 19 (2017).Crossref, Medline, Google Scholar

Vol. 9, No. 12

Supplemental Materials

Metrics

Downloaded 127 times

History

Received 11 July 2017

Accepted 11 August 2017

Published online 6 November 2017

Published in print December 2017

Information

Keywords

Acknowledgements

The authors are grateful to all the participants of the NINFEA birth cohort. The authors thank S Polidoro for her contributions to the epigenome-wide association study nested in the NINFEA cohort, and G Fiorito for his comments on an earlier version of the manuscript.

Financial & competing interest disclosure

The NINFEA study was partially funded by the Compagnia San Paolo Foundation. L Richiardi received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 733206, LIFE-CYCLE project. M Popovic was supported by the Erasmus Mundus for Western Balkans (ERAWEB II) program (reference number: E2.D2.14.270) and the present work is a part of her PhD thesis. A Biggeri was partially supported by MIUR ex 60% funds. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.

Disclaimer

This publication reflects only the author's views and the funders are not liable for any use that may be made of the information contained therein.

Ethical conduct of research

The authors state that they have obtained appropriate institutional review board approval or have followed the principles outlined in the Declaration of Helsinki for all human or animal experimental investigations. In addition, for investigations involving human subjects, informed consent has been obtained from the participants involved.

PDF download