Differentially Expressed GTEx V8 Tissues per Skipped Exon Junctions using IJC and SJC

Using rMATs.3.2.5 http://rnaseq-mats.sourceforge.net/rmats3.2.5/, we used the counts of aligned fastq files from GTEx https://www.gtexportal.org/home/tissueSummaryPage Source: GTEx Analysis Release V8 (dbGaP Accession phs000424.v8.p2). rMATS 3.2.5 discovered these skipped exon junctions by scanning Gencode human release 30 https://www.gencodegenes.org/human/release_30.html using the comprehensive GTF file. There are 42,611 distinct junctions in this skipped exon file as annotated in the fromGTF.SE.txt file.

The counts used in the differential analysis were the counts on junctions when the exon was included (IJC) counts and the counts on the junction resulting when the exon was excluded (SJC).

Also included are MDSPlots for each of the events modeled, using a linear regression model to the log2(counts + 0.05). Modeling using sex as a coefficient for the IJC and SJC models alone. And the model used for the downstream analysis, where both sex, the alternative splicing event (as_event) and their interaction (sex * as_event) were modeled. Limma's voom was used to model the variance. Random effects were used to model the effect of the donor using the block parameter to model this as a duplicate correlation event for the final voom calculation used in the fitting of the final predictive model.