A comparative analysis of cell-type adjustment methods for epigenome-wide association studies based on simulated and real data sets

• Verfasst von: Brägelmann, Johannes [VerfasserIn]
• Verfasst von: Lorenzo Bermejo, Justo [VerfasserIn]
• Titel: A comparative analysis of cell-type adjustment methods for epigenome-wide association studies based on simulated and real data sets
• Verf.angabe: Johannes Brägelmann and Justo Lorenzo Bermejo
• Jahr: 2019
• Jahr des Originals: 2018
• Umfang: 11 S.
• Fussnoten: Published: 06 August 2018 ; Gesehen am 21.04.2020
• Titel Quelle: Enthalten in: Briefings in bioinformatics
• Ort Quelle: Oxford [u.a.] : Oxford University Press, 2000
• Jahr Quelle: 2019
• Band/Heft Quelle: 20(2019), 6, Seite 2055-2065
• ISSN Quelle: 1477-4054
• DOI: doi:10.1093/bib/bby068
• Datenträger: Online-Ressource
• Sprache: eng
• K10plus-PPN: 1695277171

Abstract

Abstract: Technological advances and reduced costs of high-density methylation arrays have led to an increasing number of association studies on the possible rlationship between human disease and epigenetic variability. DNA samples fromperipheral blood or other tissue types are analyzed in epigenome-wide association studies (EWAS) to detect methylationdifferences related to a particular phenotype. Since information on the cell-type composition of the sample is generally notavailable and methylation profiles are cell-type specific, statistical methods have been developed for adjustment of cell-typeheterogeneity in EWAS.In this study we systematically compared five popular adjustment methods: the factored spectrally transformed linearmixed model (FaST-LMM-EWASher), the sparse principal component analysis algorithm ReFACTor, surrogate variableanalysis (SVA), independent SVA (ISVA) and an optimized version of SVA (SmartSVA). We used real data and applied amultilayered simulation framework to assess the type I error rate, the statistical power and the quality of estimatedmethylation differences according to major study characteristics.While all five adjustment methods improved false-positive rates compared with unadjusted analyses, FaST-LMM-EWASherresulted in the lowest type I error rate at the expense of low statistical power. SVA efficiently corrected for cell-typeheterogeneity in EWAS up to 200 cases and 200 controls, but did not control type I error rates in larger studies. Results basedon real data sets confirmed simulation findings with the strongest control of type I error rates by FaST-LMM-EWASher andSmartSVA. Overall, ReFACTor, ISVA and SmartSVA showed the best comparable statistical power, quality of estimatedmethylation differences and runtime.