Overlap of spermatogenesis-expressed genes with repetitive sequences. (A) Heatmap showing relative expression of repeats during postreplicative spermatogenesis. Heatmap is displayed according to main repeat families (labeled on left axis) and subclustered by repeat subfamilies (labeled on right axis). (B) Plot showing number of multiexonic transcripts (and percentage) not overlapping with repetitive sequences compared to monoexonic transcripts and multiexonic transcripts where repeats overlap with the first exon or any other exon. (C) Heatmap showing Z-score of Jaccard indices comparing 5′ exons to repNames at defined levels compared with randomized distributions. repNames with 10 highest Z-score values are shown. (D) Reverse empirical cumulative density functions (ECDFs) of Jaccard indices for repNames with 10 highest Z-scores (black). Transcriptome annotation was randomized 100 times to generate null distributions (red). (E) Expression level of repetitive elements (REs) associated with repNames from C across five cell types examined in this study (left) versus expression level of multiexonic transcripts containing elements of those repNames in the same data set (right). (F) Enrichment of gffcompare class codes by repName when comparing our de novo annotation to GENCODE. x-axis = log2(fold enrichment); y-axis = significance [−log10(FDR)]. (G) Genome browser depiction of the Ceacam20 locus. Top panel shows RNA-seq in five cell types with spliced reads as curved lines. Middle panel shows H3K4me3 ChIP-seq enrichment. Bottom panel shows annotation from de novo annotation, GENCODE, and Repbase. (H) Quantification of transcript levels from the Ceacam20 locus. Trans 1 represents the reference GENCODE transcript, and Trans 2 initiates from the ORR1A2 element. (I,J) Examples of genomic loci where repetitive elements lead to new transcripts lacking part of the reference GENCODE transcript. Annotation is as in G.
