Predicting disease-causing variant combinations

Abstract

Notwithstanding important advances in the context of single-variant pathogenicity identification, novel breakthroughs in discerning the origins of many rare diseases require methods able to identify more complex genetic models. We present here the Variant Combinations Pathogenicity Predictor (VarCoPP), a machine-learning approach that identifies pathogenic variant combinations in gene pairs (bi-locus variant combinations). We show that the results produced by this method are highly accurate and precise, an efficacy that is endorsed when validating the method on recently published independent disease-causing data. Confidence labels of 95% and 99% are identified, representing the probability of a bi-locus combination being a true pathogenic result, providing geneticists with rational markers to evaluate the most relevant pathogenic combinations and limit the search space and time. Finally, VarCoPP has been designed to act as an interpretable method that can provide explanations on why a bi-locus combination is predicted as pathogenic and which biological information is important for that prediction. This work provides an important new step towards the genetic understanding of rare diseases, paving the way to new clinical knowledge and improved patient care.

Usher syndrome and Nebulin‐associated myopathy in a single patient due to variants in MYO7A and NEB

“…The presence of two rare conditions, rather than one disease with phenotypic expansion, is not an unusual finding, particularly, but not exclusively, in consanguineous families. In fact, recent estimates revealed that 7% of the molecularly diagnosed patients carry a combination of two pathogenic variants implicated in distinct Mendelian disorders 10 . Therefore, care should be taken to publish new “broaden” phenotypes in well‐established syndromes particularly those that include “skeletal or muscle changes.” This report also highlights that a precise clinical characterization, and consequently a correct molecular diagnosis, might be hampered by the noncoincident age of onset of each condition, as well as adult‐onset disorders, progressive diseases, and mild phenotypes.…”

“…Homozygosity mapping proved invaluable in identifying variants in recessive Mendelian diseases, particularly in consanguineous families 11,12 . Although rare, phenotypic heterogeneity of typical Mendelian segregating disorders, caused by combined effects in several genes, has been previously described 4,10 . Yet, 44 diseases caused by 213 digenic combinations, involving 136 genes and 364 pathogenic variants, are described at the digenic diseases database (DIDA, https://omictools.com/dida-2-tool, accessed 7 May 2020) 13 .…”

Digenic inheritance of human primary microcephaly delineates centrosomal and non‐centrosomal pathways

“…New machine‐learning methods, like the variant combination pathogenicity predictor (Papadimitriou et al, ), which predicts the pathogenicity of any bilocus variant combination using a variant list from a single individual, might be used to re‐examine data from unsolved PM cases in search of variants in two genes known or suspected to interact. Putative interactions could then be validated in zebrafish.…”

“…Digenic inheritance is the simplest form of oligogenic inheritance and refers to disorders resulting from pathogenic variants at two distinct loci (Lupski, ). True digenic inheritance requires the presence of variants at two independent loci to trigger the disease, while composite class inheritance refers to Mendelizing variants with modifiers (Papadimitriou et al, ). Examples of true digenic inheritance in human pathology include facioscapulohumeral muscular dystrophy Type 2 (Lemmers et al, ) or midline craniosynostosis (Timberlake et al, ).…”

“…By contrast, dual molecular diagnosis refers to the aggregation, in a single individual, of two independent phenotypes, each caused by (a) variant(s) at a single locus (Posey et al, ). Dual diagnoses may be difficult to distinguish from digenic inheritance when the two phenotypes extensively overlap (Papadimitriou et al, ), but they are distinct from a functional point of view. Digenic inheritance, but not dual diagnoses, indicates a functional relationship between two loci, including protein–protein interaction, protein–DNA interaction, or a shared pathway, allowing for synergistic potentiation of alleles at the two loci (Schäffer, ).…”