Evaluation of Illumina 450K Methylation Chip Using Technical Replicates

• Activity Number: 649
• Type: Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: ENAR
• Author(s): Maitreyee Bose*+ and Weihua Guan and James Pankow
• Companies: University of Minnesota and University of Minnesota and University of Minnesota
• Keywords: DNA methylation ; genetic association ; intraclass correlation

Abstract

DNA methylation is an epigenetic phenomenon that may be involved in the development of diseases. In the Atherosclerosis Risk in Communities (ARIC) Study, the Illumina Infinium HumanMethylation450 (HM450) BeadChip was used to measure DNA methylation in peripheral blood obtained from ~3000 African Americans. For ~90% of the over 480,000 cytosine-guanine (CpG) sites surveyed, significant differences in methylation levels between samples on different plates was observed, largely due to technical error. To evaluate the impact of these errors, 137 replicates were used. For each CpG site, we calculated an intraclass correlation coefficient (ICC) to compare variation of methylation levels within- and between-replicate pairs; ranging between 0 and 1. Given a large number of observed ICC values at 0, we modeled the distribution of ICC as a mixture of censored or truncated normal and normal distributions using an EM algorithm. The CpG sites were clustered into low- and high-reproducibility groups, according to the calculated posterior probabilities. We then demonstrated the performance of this clustering of CpG sites on their associations with baseline characteristics like age, ethnicity etc.