Human Embryonic Expression Identifies Novel Essential Gene Candidates

Abstract

Disruption of essential genes leads to pregnancy loss, early lethality, or severe disease. Current methods to predict genes that underlie severe phenotypes include knockout animal model systems, evolutionary conservation, and variation intolerance metrics. With existing methods, human lethal genes are missed due to interspecies differences or paucity of gene characterization. We analyzed global gene expression in stages of early human development (1-cell to the blastocyst). These data were integrated with all 4049 current murine knockout phenotypes, genome-wide evolutionary gene conservation, and human genic intolerance metrics. We found that currently established human essential genes and orthologs of murine essential genes demonstrate higher gene expression across developmental stages compared to non-essential genes (Wilcoxon rank sum test, p<8.5e-10), indicating that higher expression correlates with essentiality. Of 1438 unique genes candidates with the highest expression, an estimated 1115 (78%) have not yet been associated with human disease and are thus novel candidates. The essential gene candidates concur with four prediction metrics, further supporting essentiality. We also assessed gene-specific expression changes during early development for their ability to predict essentiality. Genes that increase in expression were more likely to be essential (Fishers exact test, p<2.4e-06), suggesting that dynamic temporal expression during development may be particularly important. We find that embryonic gene expression can be used to prioritize genes that currently lack a Mendelian phenotype. Human embryonic gene expression is readily available, and applied as a novel tool, it may identify highly conserved processes vital in development.