Project Summary/Abstract Noncoding regulatory mutations had driven human evolution since the split from chimpanzees and more than 90% of disease-associated loci reside in noncoding regions. Because gene regulation is dynamic and context dependent, functions of noncoding regulatory mutations should be defined in specific cell types and at particular developmental stages. However, such a fine-resolution mapping of noncoding mutations in human diseases and evolution has been lacking in the literature, and this proposal aims to develop a comprehensive research program to close the knowledge gap by pushing our genome analysis to a single-cell resolution. My recent work has developed innovative approaches for disease genome analysis and has identified key elements driving recent human evolution. Given the prime importance of the noncoding genome in human diseases and evolution, the long-term goal of my research is to identify causal noncoding mutations that affect human phenotypes by perturbing gene regulation. Built on our recent success in capturing pathogenic noncoding somatic mutations that are predictive of prostate tumor characteristics, in the next five years, my research will push our genome analysis to a single-cell resolution. By developing a series of machine learning frameworks, we will be able to directly determine the allelic effects on altering cell-type-specific epigenomic architecture, revealing the cellular contribution to human diseases or to any evolutionary traits. Towards this goal, my laboratory is actively generating single-cell epigenome data, have obtained access to large-scale genomes for different disease categories, and have developed an innovative deep learning model achieving substantially enhanced precision for capturing pathogenic noncoding mutations. To demonstrate the general applicability of our research framework, we will investigate rare germline mutations in prostate cancer, common variants in preterm birth, and the Neanderthal introgressed alleles in the modern human genome for their regulatory effects on affecting human brain development. These conditions have wide population prevalence, and our preliminary analyses have clearly demonstrated the effectiveness of our machine learning model on identifying consequential noncoding mutations in specific cell types. We will also build an open-access genome browser which will allow users to visualize and analyze regulatory effects of any mutations in a given cell type. Overall, the proposed program will uncover new disease mechanisms from the noncoding genome, will reveal the Neanderthal impact on brain development of modern humans, and will provide a generally applicable tool for genome analysis at a single-cell resolution. Different from conventional approaches, this proposed research by introducing single-cell analysis will directly reveal the most affected cell populations in diseases, thereby guiding the future development of therapeutic strategies targeting th...