Topological Overlap Matrices for DNA Methylation data of Gestational Diabetes Cohort with BMI and Exposure Status

DNA methylation in placenta was measured with the Infinium HumanMethylation450 BeadChip (Illumina, Inc) microarray, in a sample of 28 women, 20 of whom had a gestational diabetes (GD)-affected pregnancy and 8 who did not. We used GD status as our exposure variable, assuming that this has widespread effects on DNA methylation and on its correlation patterns.  Our response, Y, is the standardized body mass index (BMI) in the offspring at the age of 5. For the 10,000 most variable probes, we provide 3 topological overlap matrices (TOM), which are used in our analysis (note that each of the following TOM matrices are a 10,000 by 10,000 symmetric matrix with row names and column names corresponding to the CpG probe IDs:

TOM_Methylation_All_10k.rds: based on all 28 subjects,  

TOM_Methylation_E0_10k.rds: based on the 8 subjects without a GD-affected pregnancy

TOM_Methylation_E1_10k.rds: based on the 20 subjects with a GD-affected pregnancy 

The BMI (phenotype) and GD status (exposure) are given in the following dataset:

BMI_and_Exposure_Status.rds: 28 x 2 matrix of the phenotype and exposure. each row is a subject.

Using our ECLUST method (preprint available at http://sahirbhatnagar.com/slides/manuscript1_SB_v4.pdf), we derive 77 clusters, and here we provide the 1st principal component of each cluster:

Cluster_Summary_1stPC.rds: 28 x 77 matrix, where each row is a subject, in the same order as the BMI_and_Exposure_Status.rds data

Cluster_CpGs_names.rds: a list of length 77, where each element of the list contains the list of CpG probe IDs contained in each of the clusters

To read in the data use the readRDS function, e.g.:

TOM_All <- readRDS(file = "TOM_Methylation_All_10k.rds")