*2.5. Bioinformatic Computation and Analysis*

The RNA sequences were mapped to hg19/GrCh37 using *HISAT2* in the Unix environment [17], and the annotated bam files were generated. The genome-wide statistical tests for differential gene

expression (DGE) between DCM patients and controls were performed using the *DESeq2* package in R programming [18]. The genome-wide statistical tests for differential exon use (DEU) between DCM patients and controls were performed using the *DEXSeq* package in R programming [19]. Features (exonic bins) with a total read count of all samples less than six were filtered out, in order to reduce false positives. The PSI score for each annotated exonic region defined by reference genome GRCh37/hg19 was computed [20]. Moreover, the methylation measurement of each probed site was also mapped to the reference genome hg19/GrCh37 through the *GenomicRanges* package in R programming [21]. Statistical tests of differential methylated regions of the 394,247 qualified probed methylation sites across the whole genome were performed with the *limma* package in R programming [22]. The M-values of DNA methylation were used. With *limma*, linear models of methylation values were defined with the following parameters: condition (DCM or control), sex, age, use of tacrolimus, use of mycophenolate, use of steroid, use of everolimus, use of ciclosporin, principal component 1, and principal component 2. Parameters sex and age were defined as categorical variables; parameters tacrolimus, mycophenolat, steroid, everolimus, and ciclosporin were binary variables, meaning intake of the specific immunosuppresive drug. Parameters principal component 1 and principal component 2 were continuous variables, which were the top two principal components of the methylation data, representing the potential substratefications of the DNA methylation data. A moderated t-statistic was applied for each probe, *p* values and adjusted *p* values in the setting of multiple testing were calculated using the Benjamini–Hochberg method. The results were mapped to the reference genome GRCh37/hg19 using the *GenomicRanges* package in R programming.
