Biobank phenotyping tools for improved understanding and management of eye diseases and disparities

PROJECT SUMMARY One of the areas of emphasis on the National Eye Institute's (NEI) strategic plan is to bridge the gap between genomics and mechanisms of disease. Vast clinically relevant information on diseases, demographics, exposures, and traits is captured in electronic health records (EHRs). This phenome data can be overlayed on biospecimen-derived data (such as genetic variants) and interrogated to uncover and clarify disease associations. These associations can in turn help us infer mechanisms of disease, contextualize clinical observations, develop risk prediction models for clinical management, and identify new research directions. Phecodes quickly capture a coded representation of the entire phenome in EHRs without accessing identifiable information, thus allowing an unrestricted exploration of any disease or phenotype as a potential cause or consequence of all others. Phecodes were last revised in 2013, when we published a replication study of known genetic associations from published GWAS. The current version of our phecodes (v1.2) has proven useful for numerous research applications and, as such, has been incorporated into novel phenome-based research pipelines to accelerate the study of both rare and complex diseases. However, our preliminary work shows that phecode v1.2's hierarchical structure lacks the resolution required to accurately identify many sigh-threatening conditions. This limits the scope of eye-related knowledge that can be gleaned from phenome-dependent association studies using biobanks. Our goal is to support new and evolving use-cases for phecodes to study eye-related diseases and health disparities in diverse biobanks. To fill this gap, we recently created a new version of phecodes (PheX) with enhanced granularity to resolve and capture many more clinical diagnoses than v1.2. We propose to: (1) Compare the resolution and reliability of PheX (vs. v1.2) to capture common and rare eye phenotypes; (2) Develop linkages between eye related-PheXs and numerous controlled vocabularies and ontologies for research; (3) Test the scalability of PheX in external biobanks through replication of eye disease GWAS results to enhance generalizability of association studies to diverse populations; (4) Incorporate PheX into novel phecode-based research pipelines; (5) Use PheX for PheWAS to study comorbid associations and to develop a catalog of morbidity and comorbidity maps with a focus on population disparities; (6) Incorporate post- coordination methods for phecode terms to enhance phenome capture; and (7) Share the new eye-specific PheX (iPheX) along with quality assessment methods in easily accessible and deployable packages to facilitate ophthalmic research in diverse biobanks. This project meets several strategic goals of the NIH and NEI, including: (1) enriching our awareness of ocular epidemiology in groups under-represented in vision science and identifying risk factors for ocular disease in underserved populations (we will use All of Us, a biobank with intentional population diversity); (2) using science to integrate cross-modality data to understand diseases (genomics, phenomics); and (3) understanding how genomic variants affect disease.