*3.3. Co-Occurrence of Di*ff*erential Exon Usage and Di*ff*erential DNA Methylation between DCM and Control*

We attempted to define regions with differential DNA methylation levels and associated differential exon usage between DCM patients and controls. This approach was set to provide important mechanistic insights into repatterning of epigenetic regulation during cardiac disease. We carried out statistical tests for differential exon use (DEU) of all exons in all gene transcripts, as well as differential methylated regions (DMR) of all probes across the whole epigenome between DCM patients and healthy controls. As mentioned in the Methods section, the *DEXSeq* package was used to perform statistical tests for differential exon use. During the process, we estimated the variability of RNA sequences data in each exonic part of each gene in each sample (Figure S4A) to effectively distinguish between actual effects across different conditions (DCM vs. control) and noises caused by biological or technical variations. Further, dispersion per exon was evaluated (black dots) and a mean relative to it was determined (rot dots) based on the estimated dispersion. Finally, the dispersion could be shrunk (blue dots) and utilized as an effective reference to examine differential exon usage. Next, statistical testing was carried out for all annotated exonic bins to determine if the fraction of the reads aligned to specific exons was different between DCM and control samples. We were able to identify 22,871 out of in total 644,354 (4%) exonic regions to be differentially used with a *p* value less than 5%. The MA plot (Figure S4B) visualized the differential exon usage based on the number of the reads mapped to each exonic region. These exonic regions were located in 8631 of the total 60,153 coding and non-coding genes (14%) annotated in hg19/GrCh37.

As mentioned in the Methods, we used the *limma* package to implement epigenome-wide statistical tests of differential methylated regions (DMR) between DCM patients and controls. As a result, we detected 13,223 of the total 394,247 (3%) probes to be significantly differentially methylated (*p* value <0.05). When overlaying the hits of DEU and DMR on the reference genome, we found 706 intron-exon pairs from 630 genes as well as 650 exon-intron pairs from 564 genes with concomitant DEU and DMR (Figure 4A). As an example of these identified candidate regions, we generated gene browser tracks for *LDB3* (Figure 5). In the gene ontology analysis, these detected genes were enriched for critical cellular components in the sarcomere, such as myofibril, contractile fibers, and actomyosin (FDR < 0.05), which are highly relevant in DCM. The relevant results of gene ontology analysis are depicted in Figure 4B, and detailed information can be found in Supplemental File 4.

**Figure 4.** Identification of genomic regions with concurrent differential exon usage (DEU) and differential methylation regions (DMR). (**A**) Intersection of DEU and DMR between DCM patients and controls. (**B**) Enrichment analysis for genes containing genomic regions with concomitant DEU and DMR.

75

**Figure 5.** Genome browser tracks demonstrating the co-occurrence of differentially used exons and the differentially methylated locus in *LDB3*. The first track represents a reference transcript, LDB3-205. The second track shows the differentially used exonic parts (green). The third track points out the position of the differentially methylated locus (red). The last two tracks are RNA-Seq coverage tracks of DCM and control.
