Infinium MethylationEPIC BeadChip FAQs

Collapse All

|

Expand All

  • Input


  • Similar to the Infinium genotyping assay, Illumina recommends DNA that is sized ≥ 2kb, which can be assessed on an agarose gel.

    Yes. Illumina recommends the following of-the-shelf bisulfite conversion kits from Zymo Research.

    • 50 reactions (single-column format) (catalog # D5001)
    • 200 reactions (single-column format) (catalog # D5002)
    • 2×96 reactions EZ-96 DNA methylation kit (deep-well format) (catalog # D5004)

    FFPE samples are already highly degraded, with a high level of crosslinking, so conversion does not occur effectively. However, you can run FFPE samples on the Infinium MethylationEPIC Array using an FFPE Methylation Protocol along with the Infinium HD FFPE DNA Restore Kit

    To run FFPE samples on the InfiniumMethylationEPIC Array, perform the following procedures:

    1. QC extract FFPE samples with the Infinium FFPE QC Kit.

    2. Perform bisulfite conversion on only samples that pass QC with one of the approved kits.

    3. Restore bisulfite-converted samples with the Infinium HD FFPE DNA Restore Kit.

    4. Input restored FFPE samples into the Infinium Methylation Assay.

  • Analysis


  • The GenomeStudio Methylation Module extracts beta values and provides clustering and normalization for Infinium MethylationEPIC array data. In addition, there are many freeware applications in Bioconductor, such as chAMP and RnBeads, which provide enhanced normalization and visualization options. Illumina does not support these freeware solutions directly.

    SNPs were included on the BeadChip so investigators could generate a DNA fingerprintof their samples as an added level of quality control. Please find further information in the Infinium HD Methylation Assay Protocol Guide. SNP assays on the BeadChip are not mentioned in the assay guide and only briefly described in the GenomeStudio Methylation Module. Follow this method to confirm the identity of samples from the same individual:

    1. Highlight the SNP assays in the sample methylation profile, right-click, and select Show only selected rows.
    2. For any given pair of samples that are supposed to be from the same individual, plot the beta values in a scatter plot from the sample methylation profile.

      Samples from the same individual:

      Infinium HumanMethylation450 samples from one individual.

      Samples from different persons:

      Infinium HumanMethylation450 samples from different people.

      The scatterplots differ because the beta values calculated from SNP assays will cluster, much as they do in a standard genotyping theta graph. Samples from the same individual have the SNP results fall along the identity line in a scatter plot, whereas samples from different persons scatter into the 9 different possible spots, based on their genotypes.

    Background subtraction is required when you compare data run on different types of scanners because of technical disparities; for example, the iScan and HiScan systems have different offsets. Background subtraction has a much smaller effect when you scan chips on the same scanner, and might not be necessary. Analyze a subset of data with and without subtraction, and choose the subset of data you prefer based on your results.

    What you are seeing are histograms of the beta values in bins of 0.02 steps and categorized by Infinium design type. The difference in beta value ranges between the Infinium I and Infinium II assay design types cause the double peaks. In general, the beta peaks at the extremes of Infinium I probes tend to be further out than the beta peaks for Infinium II.

    The beta beak differences do not affect the final analysis of the project. Individual CpG assays are not intended to be compared directly with other CpG assays, as each probe (or probe set for Infinium I designs) has different binding characteristics and behaves differently than any other probe or probe set. Rather, each assay is compared between two samples or sample groups (ie, in determining a relative rather than an absolute methylation value). For the Infinium HumanMethylation450 and Infinium MethylationEPIC array, we expect technical replicates to show > 98% correlation, and often observe > 99% correlation.