Improved filtering of DNA methylation microarray data by detection p values and its impact on downstream analyses

Clinical Epigenetics

Allan C. Just¹  Jonathan A. Heiss¹ 

[1] Department of Environmental Medicine and Public Health, Icahn School of Medicine at Mount Sinai;

关键词: DNA methylation;    Microarray analysis;    Illumina 450K;    Outlier detection;    Data cleaning;    EWAS;   

DOI  :  10.1186/s13148-019-0615-3

来源: DOAJ

【 摘 要 】

Abstract Background DNA methylation microarrays are popular for epigenome-wide association studies (EWAS), but spurious values complicate downstream analysis and threaten replication. Conventional cut-offs for detection p values for filtering out undetected probes were demonstrated in a single previous study as insufficient leading to many apparent methylation calls in samples from females in probes targeting the Y-chromosome. We present an alternative approach to calculate more accurate detection p values utilizing non-specific background fluorescence. We evaluate and compare our proposed approach of filtering observations with conventional ones by assessing the detection of Y-chromosome probes among males and females in 2755 samples from 17 studies on the 450K microarray and masking of large outliers between technical replicates and their impact downstream via an EWAS reanalysis. Results In contrast to conventional approaches, ours marks most Y-chromosome probes in females as undetected while removing a median of only 0.14% of the data per sample, catches more (30% vs. 6%) of large outliers (more than 20 percentage point difference between technical replicates), and helps to identify strong associations previously obfuscated by outliers between whole blood DNA methylation and chronological age in a well-powered EWAS (n = 729). Conclusions We provide guidance for filtering both 450K and EPIC microarrays as an essential preprocessing step to reduce spurious values. An implementation (including a function compatible with objects from the popular minfi package) was added to ewastools, an R package for comprehensive quality control of DNA methylation microarrays. Scripts to reproduce all analyses are available at doi.org/10.5281/zenodo.1443561.

【 授权许可】

Unknown