The fields of genomics and genetic epidemiology are undergoing a transformation due to two key advances: next-generation sequencing technologies and biobanks linked to electronic health records. Both these advances have led to new analytical methodologies; phenome-wide association studies that examine the impact of a genetic variant for hundreds of human phenotypes, and statistical tests that group low-frequency variants for association analyses. The current approach for grouping low-frequency variants is to consider a gene as the functional unit tested, which is counter to decades of research indicating that proteins and protein complexes are the true functional units of biology. In this project, we fuse these ideas to generate tools and analyses of ~150,000 whole-exome sequences using protein structural information from the Protein Data Bank. By performing a phenome-wide association study of low-frequency genetic variants within specific protein regions, we will construct an online atlas of functional protein regions mapped onto the clinical traits they impact.