This project aims to improve how genetic information is used to predict disease risks by validating and enhancing polygenic risk scores (PRS) through the extensive genetic and health data available in the UK Biobank. These scores are essential tools for estimating an individual’s predisposition to various diseases based on their genetic makeup.
Our approach includes transitioning from traditional relative risk measures to more clinically meaningful absolute risk scores. These provide a direct measure of an individual’s likelihood of developing a disease within a specified period, making them highly valuable for clinical decision-making. To achieve this, we will use advanced statistical models, including generalized linear models and survival analysis techniques such as Cox regression and Kaplan-Meier plots. These methods will refine how absolute risks are quantified and communicated.
Additionally, we plan to integrate whole genome sequencing data into the PRS models. This integration allows for the consideration of both common genetic variants, typically assessed in PRS, and rare pathogenic variants that significantly influence disease risk. By employing a “Rare Variant Classifier” algorithm, we will evaluate the presence of these variants and their impact on disease risk, incorporating their effects through interaction studies within generalized linear models.
To ensure the robustness and applicability of our PRS models across diverse populations, we will test their reproducibility using independent datasets from the UK Biobank and perform cross-validation studies across different ancestry groups. This rigorous validation is crucial for confirming the reliability of our risk scores.
The project is scheduled to last for 36 months, during which we will maintain close collaboration with the scientific community to ensure transparency in our methods and accessibility of our results. The goal is to provide more accurate tools for predicting disease risks, enhancing the effectiveness of medical interventions and preventive measures. This effort will not only improve individual health outcomes but also has the potential to impact public health strategies by enabling more targeted disease prevention based on genetic risk assessments.
By the end of this project, we expect to contribute significantly to the scientific literature on genetic risk prediction and support the broader use of genetic data in healthcare, enhancing public health through improved disease prevention and management.