A statistical model to identify hereditary and epigenetic fusion genes associated with dilated cardiomyopathy

Abstract

Dilated cardiomyopathy (DCM) is a heart condition that causes enlarged and weakened left ventricles and affects the heart's ability to pump blood effectively. Most genetic etiology still needs to be understood. Previously, we have used the known germline hereditary fusion genes (HFGs) to identify HFGs associated with multiple myeloma and leukemia. In this study, we have developed a statistical model to study fusion transcripts discovered from the left ventricles of 122 DCM patients and 252 GTEx (Genotype Tissue Expression) healthy controls to discover novel HFGs, ranging from 4% to 87.7%, and EFGs, ranging from 4% to 99.2%, associated with DCM. This discovery of numerous novel HFGs and EFGs associated with DCM provides first-hand evidence that DCM results from interactive developmental consequences between germline genetic and environmental abnormalities and paves the way for future research and diagnostic and therapeutic applications, instilling hope for the future of DCM treatment.

Keywords: dilated cardiomyopathy; epigenetic. RNA-Seq; fusion gene; genomics; germline; hereditary; inheritance.

FIGURE 1

Identification and analysis of fusion transcripts from DCM and GTEx. (A). Schematic diagram of a simplified procedure for identifying HFG and EFG fusion transcripts associated with DCM. The solid black and light gray rectangles represented the total fusion transcripts of the DCM patients and GTEx healthy controls; (B). Comparison of the total fusion transcripts between DCM and GTEx; (C). Comparison of numbers of fusion transcripts identified in ≥5 DCM RNA-Seq SRA samples, ≥5 DCM patients, and ≥5 GTEx healthy controls. The solid black, dark gray, and light gray rectangles showed the numbers of fusion transcripts of ≥5 DCM RNA-Seq samples, ≥5 unique DCM patients, and ≥5 GTEx healthy controls, respectively; (D). Classification and comparisons of genomic and epigenetic fusion transcripts of ≥5 unique DCM patients and ≥5 GTEx healthy controls. The solid black and gray rectangles displayed the numbers of fusion transcripts of ≥5 unique DCM patients and ≥5 GTEx healthy controls; (E). Comparisons of average frequencies of fusion transcripts of ≥5 DCM patients and ≥5 GTEx healthy controls. The solid black and gray rectangles showed the fusion transcripts of ≥5 unique DCM patients and ≥5 GTEx healthy controls; (F). The HFG, EFG, and total transcripts distribution among 122 DCM patients. The black, gray, and green lines represented the HFG, EFG, and total transcripts, respectively.

FIGURE 2

Schematic diagrams of Germline HFGs’ association with DCM. Schematic diagrams showed potential germline genomic abnormalities to generate RYR2-ACTN2. (A) NDUFV1-ACTG1 (B) and TTN-LSM1 (C). The solid black, red, and gray horizontal arrows represented the five ‘genes, 3'genes, and genes surrounded by 5′ and 3′ genes. The solid black and red squares were 5′ and 3′exons. The solid black angle line and dashed line were introns and omitted sequences. Open vertical arrows indicated potential steps from genomic alterations to producing fusion gene sequences. The numbers above the squares were exon numbers.

FIGURE 3

The heatmaps of HFG transcripts associated with DCM. (A) Morpheus generated a Germline HFG heatmap of 122 DCM patients from Broad Institute (https://software.broadinstitute.org/morpheus/). We used Morpheus to create a heatmap of 224 HFG transcripts of the 122 DCM patients. (B) Morpheus generated a heatmap of 224HFG transcripts of GTEx healthy controls from Broad Institute. K-means clustering was used to cluster rows and columns using Euclidean distance (number clustering of 2). Then, hierarchical clustering was used to cluster both rows and columns using Euclidean distance and grouping both rows and columns. The results were saved as a PDF document. The horizontal orange and light-yellow rectangles represented sparsely and highly recurrent HFG transcripts. The vertical orange and light-yellow rectangles represented HFGs-poor and HFG-rich DCM patients/GTEx healthy controls.

FIGURE 4

Schematic diagrams of potential genomic events to generate EFGs associated with DCM. Schematic diagrams showed potential mechanisms to generate EEF1DP3-FRY (A) LINC00670-MYOCD (B) and SIDT2-TAGLN (C) generated via cis-splicing of readthrough pre-mRNAs of two identical-strand neighboring genes. The solid black, red, and gray horizontal arrows represented the 5′genes, 3′gene, and genes surrounded by 5′and 3′genes. The solid black and red squares were 5′and 3′exons. The solid black angle line and dashed line were introns and omitted sequences. Open vertical arrows indicated potential steps from transcription readthrough to producing fusion gene sequences. The numbers above the squares were exon numbers.

FIGURE 5

The heatmap of EFG transcripts associated with DCM. (A). A heatmap of 181 EFGs encoding 221 EFG transcripts of 122 DCM patients was generated by Morpheus from Broad Institute (https://software.broadinstitute.org/morpheus/). We used Morpheus to create a heatmap of 221 EFG transcripts of the 122 DCM patients. (B) A heatmap of 221EFG transcripts of GTEx healthy controls was generated by Morpheus from Broad Institute. K-means clustering was used to cluster rows and columns using Euclidean distance (number clustering of 2). Then, hierarchical clustering was used to cluster both rows and columns using Euclidean distance and grouping both rows and columns. The results were saved as a PDF document. The horizontal orange and light-yellow rectangles represented sparsely and highly recurrent EFG transcripts. The vertical orange and light-yellow rectangles represented EFGs-poor and EFG-rich DCM patients/GTEx healthy controls.