Towards Reliable and Robust Causal Inference with High-dimensional Outcomes

<p dir="ltr">Tens of thousands of simultaneous statistical hypothesis tests are routinely conducted in genomic studies to identify genes causally affected by disease. Recent advances in single-cell RNA sequencing and CRISPR technologies have enabled gene expression to be measured at high resolution. However, these data are often sparse, over-dispersed, and heterogeneous, posing substantial challenges for the reliable inference of multiple cause effects.</p><p dir="ltr">This thesis develops three complementary solutions.</p>