Phenotype-Genotype Reference Map of 6,000 GWAS Associations

A new phenotype-genotype reference map pulls together genetic associations from hundreds of genome-wide association studies, and its developers say it can enable replication studies at phenome-wide scale. As they report in the American Journal of Human Genetics, researchers from Vanderbilt University and elsewhere teased out nearly 6,000 unique phenotype-genotype associations — which they limited to phecode-based phenotypes but still span 149 diseases — from 523 different GWAS to include in the phenotype-genotype reference map (PGRM). They replicated their PGRM associations in five biobank cohorts — BioVU, Michigan Genomics Initiative, UK Biobank, and BioBank Japan — reporting that 86 percent of the phenotypes replicated at least once. They applied PGRM to analyze the GWAS cohort and sniff out errors such as the accidental shifting of subject identifiers within GWAS datasets, and they further teased out variables like effect size and cohort size that affect replicability. "Replication is a powerful way to assess the data integrity of large, complex biobank datasets and to better understand analytical assumptions used in modeling phenome-wide data," the researchers write, adding that they have "created a publicly available R package that allows the PGRM to be incorporated into existing biobank QC pipelines or used to conduct replication experiments."