Adjusting for Batch Effects in DNA Methylation Microarray Data, a Lesson Learned

Abstract

It is well-known, but frequently overlooked, that low- and high-throughput molecular data may contain batch effects, i.e., systematic technical variation. Confounding of experimental batches with the variable(s) of interest is especially concerning, as a batch effect may then be interpreted as a biologically significant finding. An integral step toward reducing false discovery in molecular data analysis includes inspection for batch effects and accounting for this signal if present. In a 30-sample pilot Illumina Infinium HumanMethylation450 (450k array) experiment, we identified two sources of batch effects: row and chip. Here, we demonstrate two approaches taken to process the 450k data in which an R function, ComBat, was applied to adjust for the non-biological signal. In the “initial analysis,” the application of ComBat to an unbalanced study design resulted in 9,612 and 19,214 significant (FDR < 0.05) DNA methylation differences, despite none present prior to correction. Suspicious of this dramatic change, a “revised processing” included changes to our analysis as well as a greater number of samples, and successfully reduced batch effects without introducing false signal. Our work supports conclusions made by an article previously published in this journal: though the ultimate antidote to batch effects is thoughtful study design, every DNA methylation microarray analysis should inspect, assess and, if necessary, account for batch effects. The analysis experience presented here can serve as a reminder to the broader community to establish research questions a priori, ensure that they match with study design and encourage communication between technicians and analysts.

Genome-wide analysis of constitutional DNA methylation in familial melanoma

“…Additionally, we evaluated CpG sites with significantly lower methylation levels in familial melanoma than in healthy control samples and found 35 hypomethylated CpGs in both members of a family (Supplementary Table 2). Fifteen CpG sites showed hypomethylation in all 5 families, suggestive of a batch effect as has been described for 450k methylation arrays [20]. Of the 35 hypomethylated CpG sites, only 2 were located in established cancer-related genes: BRCA1, an established breast and ovarian cancer susceptibility gene, and ROS1, encoding a receptor tyrosine kinase with a possible oncogenic role in melanoma [21].…”

No evidence for association of MTHFR 677C>T and 1298A>C variants with placental DNA methylation

“…Raw intensity was read into Illumina Genome Studio software 2011.1, and background normalization was applied. Data processing was performed as described in Price et al [ 73 ], including sample quality checks, probe filtering, data normalization, and batch correction. This processing pipeline resulted in a final N = 442,355 cytosine-guanine dinucleotide (CpG) sites from the 450k array for analysis.…”

Transcriptome Profiling Analyses in Psoriasis: A Dynamic Contribution of Keratinocytes to the Pathogenesis

“…It is now considered to be the gold standard method for performing reproductible and accurate measurements as well as for confirming the results from high-throughput platforms [ 108 ]. Finally, another possible source of discrepancy is the batch effect, a common source of technical variation that may then be interpreted as a biologically significant finding [ 109 ]. Pairs of non-lesional and lesional skin samples might not always be processed simultaneously, thus creating a batch effect inflating the number of differently expressed genes.…”