In this paper, we demonstrate the superiority of exome sequencing data (as compared to microarray data) in researching the trait-level effects of rare variants. Using an analytical method that groups together multiple rare variants based on their co-occurrence within a gene—referred to as a gene-based collapsing analysis—we analyzed 4,264 phenotypes in 49,960 exome-sequenced individuals from the UK Biobank. We also performed replication as well as meta-analyses with 1,934 phenotypes in 21,866 members of the Healthy Nevada Project (HNP) cohort.

Our results show that 88% of the significant gene-trait associations were weakened when singleton variants—variants that appeared only once in our cohort—were excluded. Further, when variants were considered individually, 84% of genes with a significant association appeared to have no significantly associated variants. This means that the majority of the identified associations could not have been found using traditional array-based methods.

The Impact

This is the first analysis to leverage rare variant discovery and PheWAS in two large exome cohorts across thousands of medical record phenotypes to demonstrate novel biomarker discoveries. Our data show the importance of considering unique and rare variants in gene-phenotype studies. We also highlighted how the Exome+ assay can be used to discover novel genetic insights that would otherwise be missed using array-based methods.