This project develops an integrative framework – ‘PredictMix’ that can increase disease gene discovery, disease predictive and stratification accuracy, power, clinical utility, and portability in diverse ethnicities.
A key limitation of current disease risk prediction models is that they are derived using European ancestry Genome-Wide Association Studies data, making their predictive power substantially lower when computed in diverse ethnicities and mixed ancestry populations. This raises the question of how the clinical utility of these methods can be made equitable across multi-ethnic populations. On the other hand, the discovery power of GWAS is also limited in diverse and mixed ancestry populations. However, power can be optimized by combining admixture mapping and association testing, but this approach is rarely adopted because of the multi-stage process required and the challenge in application to complex multiway admixed samples. Therefore, there is a critical need to harness the power of data sciences to design an integrative framework that can increase disease gene discovery, disease predictive and stratification accuracy, power, and portability in diverse ethnicities, and has the potential to enable real-world and clinical utilities. This project is supported through The Academy of Medical Sciences – Professorship scheme for 36 months.
This project will enable a boost in power required to gain insights into heterogeneous disease etiology (Cardiometabolic trait and a range of cancers) and unlock cross-population transferability and portability issues in disease gene discovery, risk prediction, and stratification in diverse ethnicities.
