Improving genetic risk prediction across diverse population by disentangling ancestry representations
This addresses the bias in polygenic risk models that affects minority populations and admixed individuals, representing a novel method for a known bottleneck rather than an incremental improvement.
The paper tackled the problem of genetic risk prediction models performing poorly for minority populations due to biases from European ancestry data, and the result was a novel deep-learning framework that disentangles ancestry from phenotype-relevant information, substantially improving risk prediction in minority populations, particularly for admixed individuals, as demonstrated in Alzheimer's disease genetics analysis.
Risk prediction models using genetic data have seen increasing traction in genomics. However, most of the polygenic risk models were developed using data from participants with similar (mostly European) ancestry. This can lead to biases in the risk predictors resulting in poor generalization when applied to minority populations and admixed individuals such as African Americans. To address this bias, largely due to the prediction models being confounded by the underlying population structure, we propose a novel deep-learning framework that leverages data from diverse population and disentangles ancestry from the phenotype-relevant information in its representation. The ancestry disentangled representation can be used to build risk predictors that perform better across minority populations. We applied the proposed method to the analysis of Alzheimer's disease genetics. Comparing with standard linear and nonlinear risk prediction methods, the proposed method substantially improves risk prediction in minority populations, particularly for admixed individuals.