Project Summary/Abstract Differential local adaptations in human populations during our evolutionary history have contributed to the variation in phenotypes and disease risks observed in modern humans. While statistical genomic signatures of local adaptation can be detected across ancestries, the variants and genes underlying these adaptive traits, as well as their function, remain nearly entirely unknown. The signature of high linkage disequilibrium (LD) that enables detection of adaptive loci also hinders our statistical fine-mapping ability at these loci. Moreover, these variants are often in non-coding, cis-regulatory elements (CREs) of the genome, in which it is difficult to predict the impact of genotype on function. Additionally, current GWAS and quantitative trait studies commonly used to interpret regulatory variation, are European-biased, rendering them underpowered in diverse ancestries. Discovering the evolutionary trends, mechanisms, and phenotypes of adaptive variation therefore requires LD-independent, ancestry-agnostic tools that can characterize putatively adaptive variation at scale. During my dissertation I will generate a catalog of well-characterized putatively-causal adaptive CRE variants by pairing previously-designed machine-learning detectors of adaptation with scalable functional genomics technologies. DeepSweep, a machine-learning algorithm developed by the Reilly Lab, drastically reduces the number of putatively causal variants per causal loci to the point that they can be experimentally tested. I will functionally fine-map DeepSweep-identified putatively adaptive variants from non-admixed 1000 Genome Project populations using the Massively Parallel Reporter Assay to identify expression-modulating variants (emVars). I will next identify the gene targets of functional adaptive variants using single-cell CRISPR-interference screens. Knowing these gene regulatory targets will enable an understanding of the biological processes, pathways and higher-order phenotypes impacted by adaptation. This catalog of functionally fine-mapped adaptive CRE variants linked to target genes will resolve phenotype impacting variation at high-LD loci, and for the first time, provide enough instances of characterized adaptation to discover trends of adaptation shared across ancestral histories. I will Interrogate the medical utility of this catalog by intersecting emVars with phylogenetic constraint scores and emVar regulatory targets with genome-wide association catalogs and gene ontologies. The prevalence of recurrent adaptation in human evolution will be investigated by intersecting emVars with transcription factor motif databases and pathway analysis of emVar regulatory targets. Furthermore this catalog of global adaptive variation will be increasingly useful in resolving causal trait-associated variants at high-LD adaptive loci as association studies become more global and diversified.