Significant effort has been dedicated to the study of complex diseases through genome-wide association studies (GWASs). However, GWASs have provided few medically actionable results due to several limitations. One limitation of the GWAS approach is the difficulty in identifying functional variants when most assayed SNPs have no known function and/or are considered tags for an ungenotyped or uncharacterized functional variant. A second limitation is the lack of heritability or appreciation for the environmental contribution to the disease under study. Alternative and complementary approaches to the GWAS technique are necessary. A novel strategy to address these limitations includes the use of electronic medical records (EMRs) to conduct a phenome-wide association study (PheWAS) when plausible genetic targets are identified. Whereas GWAS asks What genetic variants are associated with a disease? PheWAS asks What diseases are associated with a genetic variant? The hypothesis being tested in this project is that loss-of-function variants - a class o variation with the highest probability of being clinically relevant - may cause disease phenotypes described in EMRs. To test this hypothesis, we propose the following specific aims: (1) measure associations between thousands of disease phenotypes coded in the EMR for 10,000 Marshfield Clinic patients and loss-of-function SNPs, (2) replicate findings in an independent cohort, and (3) investigate the biological relevance of these associations through functional genomic experiments using patient biospecimens, cell lines, and animal model systems. The results from this study will combine association-based testing with biological experimentation. Therefore, this study is not only innovative in its approach, but also in its capacity to assess the genetic component of many diseases simultaneously, including understudied diseases. In addition, the PheWAS approach has the capacity to characterize multiple diseases that share a common genetic etiology. This may be important for drug repurposing, in that drugs used to treat one disease may also be therapeutic for a different disease, if both share a common genetic link.
Alternative approaches are needed to efficiently identify clinically actionable genetic variants for use in personalized medicine. One such approach includes linking genetic data with in-depth patient data from electronic medical records that describe many diseases simultaneously, coupled with experiments that are designed to understand the biology of the genes and diseases. This strategy may be useful not only in predictive medicine, but may also provide new therapies by expanding treatment options for existing drugs impacting biological pathways shared by multiple diseases.
Showing the most recent 10 out of 12 publications