Leveraging Electronic Health Record Derived Phenome to Enhance Genetic Studies of Common and Rare Disease
Abstract
The integration of electronic health record (EHR) derived phenotypes with genetic data has been instrumental in enhancing our understanding of gene-phenotype associations. Vanderbilt University Medical Center has an extensive EHR system that has over 3.5 million individuals that are stored in a de-identified database, the synthetic derivative (SD) that includes ICD billing codes, medications, clinical notes, medical histories, and laboratory values. A subset of the SD, BioVU, is a de-identified EHR-linked DNA repository. Having EHR-linked genetic data has allowed for the development of scalable and accurate phenotyping methods (phecodes), the expansion of analyses beyond genome-wide association studies (GWAS) to identify associations phenome, transcriptome, and laboratory-wide (PheWAS, TWAS, LabWAS). Additionally, methods such as the development of phenotype risk scores (PheRS) have allowed us to disentangle distinct phenotypic patterns of disease and clinical presentation of the complex, multisystem Mendelian diseases as well as enhance our understanding of disease comorbidities. Leveraging these data and methods, we have been able to study common genetic disorders and provide biological context to shared genetic architecture of eye disease by implementing gene and comorbidity-based analyses. These analyses were able to identify established and novel eye disease associations. Additionally, these data and resources have allowed us to study Mendelian diseases on a large scale. We implemented an approach that leverages PheRS to ask questions about Mendelian disease gene function and explore gene-phenotype associations among genes used for clinical diagnostic testing with accurate phenotyping of craniofacial congenital anomalies. The ability to implement these methods to ask questions about both rare and common genetic diseases highlights the diversity and breadth of research questions that can be asked of EHR-linked genetic data.
Description
Keywords
electronic health records, genetics, biobank, TWAS, PheWAS, PheRS