PROJECT SUMMARY Recurrent pregnancy loss (RPL) affects up to 5% of couples, yet nearly half of cases remain unexplained by current testing recommendations. Euploid pregnancy loss, in the setting of unexplained RPL, is particularly frustrating for patients and providers because there is no clear explanation or any proven therapies to mitigate risk of subsequent miscarriages. As clinical presentation and subsequent pregnancy outcomes vary widely, this complex disorder will ultimately require a precision health approach. While more than 3000 human genes are conserved and likely essential for early development, remarkably little is known about their contribution to RPL and current genetic databases are essentially devoid of RPL entries. Moreover, there is currently no database that annotates phenotypes and genotypes of these essential genes. This proposal aims to define genetic determinants of RPL through clinical and molecular phenotyping and genomic sequencing of a large RPL cohort, combined with novel bioinformatics and machine learning approaches to derive predictive risk algorithms. A comprehensive approach to identify genomic markers of pregnancy loss by whole genome sequencing of well- characterized RPL trios (mother-father-pregnancy loss) will be undertaken in Aim 1. These genetics efforts will be paired in Aim 2 with metabolomic, lipidomic and single cell transcriptomic profiling preconception and in early pregnancy. Leveraged with innovative machine learning strategies in Aim 3, this approach will significantly advance understanding of the genetic underpinnings of unexplained RPL. A clinical ‘intolerome’ database will be constructed in Aim 4 to facilitate worldwide collaboration and curation of genotypes and associated phenotypes, making the genetics and omics data and results available to the public as well as other funded teams. This multidisciplinary team includes leaders in RPL, genetics, genomics, prenatal diagnosis, bioinformatics and machine learning at Stanford, UCSF and OHSU. Combined we have a substantial cohort of RPL patients that will serve as a robust recruitment source, along with a collaboration with the unique UK Pregnancy Baby BioBank of existing trios to accomplish project goals. The proposed study is anticipated to have significant clinical and research impact by identifying the genomic contribution to RPL in a large and well phenotyped cohort and building improved risk predictions based on machine learning incorporating clinical, genetic, and molecular data. This work will lay the foundation for precision medicine-based interventions for RPL couples who are difficult to diagnose and have few proven treatments.