The impact of rare variation on extreme and unexpected phenotypes in the UK Biobank

Thesis

Human disease can be characterised by deviation from a ``normal'' or healthy state. Using exome sequencing data from the UK Biobank (UKB), I examined the effect of rare coding variants on deviations from normality to better understand disease. Biomarker deviations from reference ranges are commonly used to indicate disease. For the first time, I quantified the proportion of individuals in UKB with abnormal biomarker values across 32 blood and urine biomarkers. Disease risk increased with the number of abnormal biomarkers, highlighting their clinical relevance. By contrasting traditional linear association between continuous biomarker level with dichotomised (extreme vs. non-extreme value) biomarkers, I identified associations which would not otherwise have been found, including a PNPLA3 missense variant associated with extreme serum triglyceride levels. Polygenic scores (PGS) predict complex traits and stratify disease risk but fail to fully capture individual-level variation. Misaligned individuals, whose phenotypes deviate from expectation based on PGS, allow us to identify risk factors beyond common-variant effects. I tested whether rare damaging variation in rare disorder genes can explain this deviation for ten traits: type 2 diabetes (T2D) cases carrying pathogenic variants have significantly lower PGS than non-carriers, and both coronary artery disease and T2D controls carrying protective variants have significantly higher PGS than non-carriers. Using this framework, I searched for novel protective or pathogenic genes, identifying KANK1 as a therapeutic target for primary ovarian insufficiency. Finally, I led the largest-ever systematic functional characterisation of genes associated with fat accumulation, a disease mechanism in which abnormal deviations cause obesity and lipodystrophy. I identified SLTM as a novel therapeutic target for obesity, due to its effect on earlier onset of obesity, its replication in an independent dataset (All of Us), and the significant effect of SLTM knockdown on lipid accumulation in human fat cells. Taken together, this thesis advances understanding of the impact of rare variants on extreme and unexpected phenotypic deviations, and contributes meaningful insights into the genetic aetiology of human disease.

Authors: Baya, NA

• DOI: 10.5287/ora-komzxxp2n
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: predicted loss-of-function; CRISPR; therapeutic target discovery; Familial hypercholesterolaemia (FH); adipocytes; Mahalanobis distance; osteoporosis; phenotypic deviation; monogenic diabetes; rare variant analysis; polygenic scores; reference range outliers; statistical genetics; exome sequencing; Coronary Artery Disease; human disease; obesity; Maturity-onset diabetes of the young (MODY); genomic medicine; outliers; biomarkers; burden testing; precision medicine; Diabetes; Lipodystrophy; primary ovarian insufficiency; UK Biobank
• Subjects: Obesity; Statistics; Genetics; Biochemical markers; Reference values (Medicine); Biometry; Genomic medicine