*2.7. Methylome-Transcriptome Correlation*

To further investigate the methylome–transcriptome correlation in DCM, we accomplished an epigenome-wide association analysis, modified from a genome-wide association study (GWAS), between the DNA methylation measurement (beta value) and PSI score. The correlational analyses were done both in DCM patients and in control subjects. The advantage of this approach was that not only local but also remote regulation of intronic DNA methylation on the exon inclusion within the same gene could be thoroughly explored. The single locus analysis was used. Logistic regression was applied as it permits adjustment with additional parameters and provides odds ratios as effect sizes [25]. An odds ratio and a *p* value were calculated for each generated regression model. The first two principal components were integrated in the regression model as covariates in order to account for possible population stratification and to minimize the inflation. The robustness of this genome-wide association approach was evaluated and subsequently optimized following the criteria suggested commonly [26]. In addition to the above-mentioned procedure, we carried out genome-wide statistical tests of significance for the difference between correlation coefficients in DCM patients and in controls. The aim of this approach was to identify genomic regions with significantly divergent methylome-transcriptome relationship between DCM patients and controls. The statistical tests were performed using Fischer's z test with the help of the *cocor* package in R programming [27]. The significance threshold was determined to be FDR < 0.05 by Benjamini–Hochberg in the screening cohort and a raw *p* value < 0.05 in the replication cohort.
