Guidance on filtering DNA methylation microarray probes by detection p-values

DNA methylation microarrays are popular for epigenome-wide association studies, but spurious outlier 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 probes on the Y-chromosome in samples from females. We extend these observations by assessing 2,578 samples from 18 studies on the 450K microarray as well as outliers between technical replicates. We present an alternative and corrected detection p-value calculation that utilizes an out-of-band intensity estimation of the background signal. This approach catches 63% of large outliers (more than 20 percentage points between replicates) versus 9% with conventional approaches. We provide comprehensive guidance and an R package ewastools for filtering both 450K and EPIC microarrays as an essential preprocessing step to reduce outliers.