There is a disconnect between the study of the role of common genetic variation and the study of rare protein coding variants in human disease. From a natural selection point of view, there is no common or rare, but just a sequential set of genetic variation that has been generated, selected for or against competitive advantages.
We are aiming to use the UK Biobank data to predict those individuals with a high risk for common and rare human conditions. We will combine millions of common genetic variants in the form of polygenic scores with rare protein coding variants. We expect that this research will help us understand why some relatives of patients with mendelian diseases do not get sick despite also carrying the same pathogenic variant as the proband.
To accomplish this goal, we will combine genotypic and phenotypic data from the UK Biobank dataset to build polygenic models that will then be used to predict individual phenotypes. External resources such as summary statistics from GWAS meta-analysis, large-scale RNA-seq, proteomics, or other population genetics studies will be leveraged to differentiate functional vs natural variation. For the analysis of the genetic data, we will use a combination of the imputed data as well as the exome-sequence data from jointly-called VCF files.