The goal of the proposed project is to develop and apply statistical methods to better predict the individual risk of patients for a specific disease based on genetic data.
Current attempts to predict certain traits or phenotypes such as e.g. diseases or physical characteristics based on individual genomic data lack to detect all of the expected heritability. This might partly be due to the use of simplified approaches to model the association of genotypic and phenotypic data. For example polygenic risk scores (PRS), which measure the individual genetic risk for a certain trait, are usually constructed considering each genomic variant independently and not by analyzing all variants simultaneously.
Finding more accurate models with higher prediction ability can help to tailor treatments for the individual patient and to improve preventive measures. Invasive therapies might be avoided for people with low risk and appropriate preventive measures can be taken for high risk patients. The aim of the project is to develop those integrative and flexible models combining different sources of information. Building such disease risk and stratification algorithms requires a significant amount of resources and effort. Hence, we believe that a duration of 36 months is required to achieve the aims of this project.
The algorithms will be implemented in tools which will be made available as open-source software for potential application in the biomedical field (e.g., disease risk stratification).