Abstract
The biggest challenge geneticists face when applying next-generation sequencing technology to the diagnosis of rare diseases is determining which rare variants, from the dozens or hundreds detected, are potentially implicated in the patient's phenotype. Thus, variant prioritization is an essential step in the process of rare disease diagnosis. In addition to conducting the usual in-silico analyses to predict variant pathogenicity (based on nucleotide/amino-acid conservation and the differences between the physicochemical features of the amino-acid change), three important concepts should be borne in mind. The first is the "mutation tolerance" of the genes in which variants are located. This describes the susceptibility of a given gene to any functional mutation and depends on the strength of purifying selection acting against it. The second is the "mutational architecture" of each gene. This describes the type and location of mutations previously identified in the gene, and their association with different phenotypes or degrees of severity. The third is the mode of inheritance (inherited vs. de novo) of the variants detected. Here, we discuss the importance of each of these concepts for variant prioritization in the diagnosis of rare diseases. Using real data, we show how genes, rather than variants, can be prioritized by calculating a gene-specific mutation tolerance score. We also illustrate the influence of mutational architecture on variant prioritization using five paradigmatic examples. Finally, we discuss the importance of familial variant analysis as final step in variant prioritization.