A genome-wide case-only test for the detection of digenic inheritance in human exomes

Abstract

Whole-exome sequencing (WES) has facilitated the discovery of genetic lesions underlying monogenic disorders. Incomplete penetrance and variable expressivity suggest a contribution of additional genetic lesions to clinical manifestations and outcome. Some monogenic disorders may therefore actually be digenic. However, only a few digenic disorders have been reported, all discovered by candidate gene approaches applied to at least one locus. We propose here a novel two-locus genome-wide test for detecting digenic inheritance in WES data. This approach uses the gene as the unit of analysis and tests all pairs of genes to detect pairwise gene x gene interactions underlying disease. It is a case-only method, which has several advantages over classic case-control tests, in particular by avoiding recruitment and bias of controls. Our simulation studies based on real WES data identified two major sources of type I error inflation in this case-only test: linkage disequilibrium and population stratification. Both were corrected by specific procedures. Moreover, our case-only approach is more powerful than the corresponding case-control test for detecting digenic interactions in various population stratification scenarios. Finally, we validated our unbiased, genome-wide approach by successfully identifying a previously reported digenic lesion in patients with craniosynostosis. Our case-only test is a powerful and timely tool for detecting digenic inheritance in WES data from patients.Significance statement. Despite a growing number of reports of rare disorders not fully explained by monogenic lesions, digenic inheritance has been reported for only 54 diseases to date. The very few existing methods for detecting gene x gene interactions from nextgeneration sequencing data were generally studied in rare-variant association studies with limited simulation analyses for short genomic regions, under a case-control design. We describe the first case-only approach designed specifically to search for digenic inheritance, which avoids recruitment and bias related to controls. We show, through both extensive 4 simulation studies on real WES datasets and application to a real example of craniosynostosis, that our method is robust and powerful for the genome-wide identification of digenic lesions. 5

Centers for Mendelian Genomics: A decade of facilitating gene discovery

“…50 Reanalysis methods for more complicated patterns of inheritance led to the development of a two-locus genome-wide test that enables detection of digenic inheritance in exome data. 51 CMGs investigators have successfully applied sequencing approaches beyond the exome to identify or validate causal variant(s), including genome, RNA, bisulfite DNA methylation sequencing, and long-read sequencing, showing their utility in cases where exome sequencing fails to find a molecular diagnosis. Examples include utilizing genome sequencing to identify pathogenic SVs missed by exome, such as the homozygous inversion in QDPR detected in a patient with dihydropteridine reductase deficiency; 52 applying RNA sequencing to identify genes with aberrant expression and or splicing, including an intronic variant in trans with a missense in muscle disease gene DES that resulted in a pseudo-exon insertion and allelic imbalance; 53 using bisulfite sequencing to identify gene silencing epivariation, such as the characterization of aberrant hypermethylation associated with a pathogenic repeat expansion in the XYLT1 promoter region; 54 and the application of long-read sequencing to characterize a complex genomic rearrangement involving an inverted triplication flanked by duplications in a proband with Temple syndrome.…”

“…50 Reanalysis methods for more complicated patterns of inheritance led to the development of a two-locus genome-wide test that enables detection of digenic inheritance in exome data. 51…”

The GENDULF algorithm: mining transcriptomics to uncover modifier genes for monogenic diseases

“…Points are defined as outliers if they are greater than q 3 + w × ( q 3 − q 1 ) or < q 1 − w × ( q 3 − q 1 ), where w is the maximum whisker length, and q 1 and q 3 are the 25 th and 75 th percentiles of the sample data, respectively. There are five severe (red), four mild (orange), and eleven control (blue) biological replicates. DMap of coincidence of the low expression (lowest 10% of expression levels) of the 7 top DPMs with the low expression of CFTR in GTEx lung samples (lowest 10% of expression levels, when symptoms would be present in CF patients Ramalho et al , 2002; Kerem et al , 1997); each small rectangle inside the big rectangle represents one individual; all presented samples are those with low CFTR expression. Dark blue rectangles indicate samples with low expression of the listed DPM). E, FUpper panels: scatter plot associating the expression of the GCD (CFTR) vs. identified PMs in healthy colon GTEx tissue; the expression of CFTR ( x ‐axis) vs. that of (E) CLCA1 and (F) SLC4A4 , respectively, in healthy colon tissues.…”

“…While there is growing recognition that genomic context strongly modulates the clinical severity of many monogenic disorders, few modifier genes have been identified to date (Kousi & Katsanis, 2015). Recently, Kerner et al (2020) proposed a case‐only, disease‐specific, genome‐wide strategy based on DNA variants identified in whole exome sequencing. One can also search simultaneously for modifiers of many disease genes in large sequencing databases such as ExAC by identifying putatively healthy individuals who carry one or more variants that are expected to cause a monogenic disorder (Tarailo‐Graovac et al , 2017); this approach also uses DNA analysis (not gene expression) and is not targeted to a disease since databases such as ExAC ascertain healthy individuals.…”

Oligogenic combinations of rare variants influence specific phenotypes in complex disorders

“…In addition, the common practice of evaluating feature importance metrics of machine learning classifiers falls short of the objective to identify combinations of features that exert higher effect on the phenotype than evident from their independent effects 17,18 . Furthermore, prior studies to assess combinatorial effects have been inherently biased due to their need to minimize the search space by restricting the analysis to only a subset of genes chosen based on a priori knowledge [41][42][43] . Here, we provide a proof-of-concept analytical framework that remains agnostic to prior evidence and performs exhaustive searches to identify combinatorial effects among rare variants while retaining high granularity of data and interpretability of results.…”