A major goal in genetics is to use genetic information to predict phenotype, which could have enormous impact in medicine and in our general understanding of biology. We are contributing towards this goal by using high-throughput genome engineering methods to generate and study thousands of genetic variants. This will allow us to understand the consequences of the specific variants we study, and also learn about the general principles underlying variant consequences for extension to the even broader space of untested variants. To accomplish high-throughput genome engineering, we use large-scale parallelized oligonucleotide synthesis to generate pools of thousands of unique yeast plasmids, each of which carries a guide RNA (gRNA) gene and a paired repair template encoding a specific mutation of interest. Following directed DNA cleavage by Cas9 using the gRNA, the repair template introduces the mutation in the yeast genome. The edited pool of yeast cells is then subjected to selection for a phenotype of interest. The abundances of the unique plasmids in the selected pool thus reports on the effect each genetic variant had on the phenotype under study. We are also developing new high-throughput genome engineering approaches. These include advancements to CRISPR technology, by expanding the space of genetic outcomes that CRISPR can target, as well as novel applications of high-throughput editing, such as careful deployment of CRISPR-directed deletions. In the past year, we contributed to a study of the genetics of human height in Jewish families. This study was designed to test the hypothesis that despite the majority of heritability in human height coming from common genetic variants, rare variants nonetheless might often have large phenotypic consequences, with this overall contribution to heritability being masked by such variants' rarity. By specifically looking at the genetics of height within families, rare variants will segregate at the same frequency as common variants, and thus quantitative trait loci (QTLs) uncovered in these families could have larger effect sizes than seen through genome-wide association studies on unrelated individuals. Indeed, this study found evidence of large-effect QTLs segregating in the families under study, supporting the initial hypothesis.
