Up to 80% of rare disease patients remain undiagnosed after genomicsequencing1, with many probably involving pathogenic variantsin yet to be discovered disease–gene associations. To search for suchassociations, we developed a rare variant gene burden analytical framework forMendelian diseases, and applied it to protein-coding variants from whole-genomesequencing of 34,851 cases and their family members recruited to the 100,000 GenomesProject2. A total of 141 new associations were identified,including five for which independent disease–gene evidence was recentlypublished. Following in silico triaging and clinical expert review, 69 associationswere prioritized, of which 30 could be linked to existing experimental evidence. Thefive associations with strongest overall genetic and experimental evidence weremonogenic diabetes with the known β cell regulator3,4UNC13A, schizophrenia with GPR17, epilepsy with RBFOX3, Charcot–Marie–Tooth disease with ARPC3 and anterior segment ocular abnormalities withPOMK. Further confirmation of these and otherassociations could lead to numerous diagnoses, highlighting the clinical impact oflarge-scale statistical approaches to rare disease–gene associationdiscovery.
