18083
Analysis of Differential Methylation in Autism Spectrum Disease Using a Novel Probe-Based Algorithm
Hereditary origins for Autism Spectrum Disorder are well established, but other factors are also involved in ASD, as demonstrated by monozygotic twins discordant for autism traits. We studied the role of epigenetics in autism, by comparing methylation profiles of blood samples of 42 ASD samples vs. 108 controls, using the 27K Illumina platform.
Objectives:
To identify probes showing differential methylation in individuals with ASD vs. controls.
Methods:
Methylation studies using the Illumina platform are based on the metric Beta, with states of methylation assigned with sample-based algorithms. We have observed that these approaches assign wrong methylation states to many probes, which led us to develop a probe-based algorithm. We analysed changes in states of methylation/hemi-methylation/unmethylation in ASD samples vs. controls, which we believe is a realistic representation of the underlying biology of the epigenetics of ASD. We have used data mining techniques, such as the PAM clustering algorithm to identify probes with similar profiles. The statistical package R and Python were used in our analysis.
Results:
We have identified 40 probes that show differentially methylation in cases vs. controls, including CpG-islands close to the genes EPHA5, CDK5 and WHSC2. All our samples are of Irish origin; we performed ancestry analysis to validate ancestry was not a counfounding factor in differential methylation. We apply our methodology to a dataset of concordant and discordant ASD twin samples, which is publicly available (Wong, 2013); we observe little overlap between the two datasets, but a few probes show similar methylation profiles.
Conclusions:
We observe differential methylation for 40 probes in a cohort of 42 Irish individuals with ASD. Some of the probes are close to genes that might have a role in ASD, such as the axon guidance gene EPHA5. Our study suggests a role for epigenetics in ASD.