PROJECT SUMMARY
Genome-wide association studies have now discovered tens of thousands of noncoding variants associated with
human diseases and traits. It has proven challenging to interpret these associations. A majority of causal variants
lie in the noncoding genome and appear to affect DNA cis-regulatory elements, which control the logic of gene
expression and could point us to new cell types, genes, and pathways for disease. However, we have lacked
the tools needed to systematically characterize how these cis-regulatory variants and elements impact genome
function and phenotype.
Our team at Stanford University has now developed innovative single-cell, CRISPR mapping, and computational
technologies that will enable identifying and functionally characterizing many thousands of elements and variants
directly in the human genome. These tools include single-cell ATAC-seq to identify candidate elements in cells
and tissues; sensitive CRISPR tiling methods to connect thousands of elements and variants to effects on gene
expression and cellular phenotypes; and the ABC and BPNet models to predict how disease variants regulate
gene expression. Together, these technologies suggest a new strategy to systematically connect DNA variants
and elements to function and phenotype.
Here we will apply these new technologies in collaboration with the NHGRI Impact of Genomic Variation on
Function Consortium. We will use four cardiovascular cell types derived from human pluripotent stem cells as
model systems. First, we will leverage single-cell maps of cardiac differentiation and development to select
elements and risk variants for adult and children’s heart diseases likely to control cardiovascular cell function.
Second, we will apply single-cell CRISPR tools to measure the effects of thousands of unbiased elements and
variants on gene expression, and connect prioritized disease variants to target genes, cellular phenotypes, and
tissue phenotypes. Third, we will leverage these experimental datasets to calibrate and refine computational
models to build a variant-element-phenotype catalog across many human cell types and diseases. Fourth, we
will enable future studies by sharing data, protocols, and software, and by conducting systematic evaluations of
CRISPR technologies and computational models to connect variants to phenotypes. Together, these studies will
advance our understanding of how DNA variants and elements impact genome function and demonstrate a novel
strategy to leverage high-throughput genomic tools to understand biological mechanisms of human diseases.