DISSECTING THE POLYGENIC LANDSCAPE OF HUMAN COMPLEX TRAITS AND DISEASES IN A SOUTHERN ITALIAN COHORT

Complex traits and diseases are shaped by the combined effects of common and rare genetic variants, yet their contribution to risk is still poorly characterised in Southern European populations. This thesis addresses this gap by analysing the genetic architecture of traits and diseases in Italy, leveraging the Moli-sani cohort, a large population-based study from Southern Italy. The aim is to clarify the role of genetic variation in disease risk and to evaluate its potential for improving prediction models with applications in precision medicine and public health. In the first part, genome-wide association analyses of quantitative traits showed overall consistency with findings from other European populations, while also revealing population-specific features, such as the persistence of strong-effect alleles at the HBB locus. We then assessed the transferability of polygenic scores derived from Northern European datasets to the Italian context, finding that predictive accuracy was reduced for several traits. However, incorporating variants specific to the Italian population improved performance in some cases, highlighting the value of population-tailored models. The second part explored the predictive potential of polygenic risk scores (PRS) for major diseases using long-term follow-up and electronic health records. PRS stratified lifetime risk for conditions such as coronary heart disease and breast cancer, and when combined with demographic factors such as age and sex, achieved performance comparable to SCORE2, the standard clinical risk model. Unlike traditional risk factors, PRS are fixed at birth and can be calculated early in life, enabling lifelong risk stratification and supporting prevention strategies that can be tailored to the individual. The final part of this thesis focused on rare variants identified through whole-genome sequencing. Gene-based analyses of 52 quantitative traits yielded 95 significant associations, approximately one third of which were replicated in the GeneBass resource. A leave-one-variant-out analysis further prioritised 238 alleles likely driving these signals, including rs730882080 (FH Naples-3) in LDLR, a pathogenic variant first reported in Southern Italy and associated with elevated LDL cholesterol levels and an increased risk of coronary heart disease. These findings demonstrate the value of studying underrepresented populations to characterise their unique genetic architecture, improve the accuracy of genomic prediction, and identify variants with direct clinical relevance. By integrating analyses of common and rare variants, polygenic scores, and sequencing data, this thesis contributes to the foundations of precision prevention and highlights the potential of genomics to inform public health in Italy.