A Review of the Illumina Infinium Humanmethylation450 Beadchip Assay Design and Implications for Normalization

• Activity Number: 194
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Cong Lu*+ and Diane E. Grill and Douglas W. Mahoney and Gregory A. Poland and Ann L. Oberg
• Companies: Carnegie Mellon University and Division of Biomedical Statistics, Division of Health Sciences Research, Mayo Clinic and Division of Biomedical Statistics, Division of Health Sciences Research, Mayo Clinic and Mayo Clinic Vaccine Research Group, Mayo Clinic and Division of Biomedical Statistics, Division of Health Sciences Research, Mayo Clinic
• Keywords: DNA methylation ; Normalization ; HumanMethylation450

Abstract

The role of DNA methylation in biologic and disease processes is of great scientific interest. The Illumina Infinium HumanMethylation450 BeadChip is a microarray assay that interrogates DNA methylation at over 485,000 CpG sites in the human genome. Two bead designs are used in this microarray. In one, each CpG site is represented by two different colored beads. In the second, each CpG site is represented by one bead containing two colors. The distribution of expression values is known to differ between the two designs as a function of the number of CpGs in a probe. In data from 150 human subjects' blood, we demonstrate evidence of non-linear between-subject biases that differ between the methylated and unmethylated probes. But between-specimen biases appear to be linear on both the beta-value (i.e., percent methylation) and M-value scales (i.e., the logit of the beta-values). While application of two common normalization strategies result in a similar distribution of between-specimen variance across the probes, estimates of percent commonly vary by 10-20%. We discuss implications of the design and observed biases for both the within-array and between-array normalization strategies.