High-dimensional Statistical Genetic Approach for Family-based Orofacial Clefts

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Although family studies were the basis for genetic risk prediction before the advent of modern molecular markers, they have been much less developed for risk prediction of complex diseases using high-dimensional data. Family studies offer many ideal features for large-scale risk prediction research. It provides robust protection against confounding bias when dealing with samples from multiple ethnic groups (i.e., population stratification). Aside from that, Family studies could take into account family information (i.e., genotype and phenotype information from family members) for improved risk prediction. Despite these advantages, they have been used infrequently in recent risk prediction research. The goals of this application are to develop a statistical genetic approach for high-dimensional family-based risk prediction, and to build a family-based risk prediction model by applying the proposed approach to the International Consortium of Orofacial Clefts genome-wide association study dataset. The central hypothesis is that the proposed approach, which considers a large number of genetic and environmental predictors, family information and population substructure, will outperform an existing generalized estimating equations based genotype scoring approach (GEE-GS), and will lead to a robust and accurate family-based risk prediction model for orofacial clefts. The proposed research will be initiated by an early-stage new investigator, who has assembled a research team of senior scientists, including Robert C. Elston, Jeffrey C. Murray and Brian Schutte. The team has developed novel statistical genetic approaches for risk prediction research, and has been active in orofacial clefts genetic and clinical research. In the proposed research project, the research team will turn its attention to family-based orofacial clefts risk prediction. The planned specific aims are to: 1) Develop a robust clustered likelihood ratio approach for high-dimensional family-based risk prediction and compare its performance with the GEE-GS approach through extensive simulation studies; and 2) Build a high-dimensional orofacial clefts risk prediction model by simultaneously considering a large number of genetic and environmental predictors, their interactions, and family information. If successful, the new approach will facilitate high-dimensional family- based risk prediction studies in general. The orofacial clefts risk prediction study will also lead to a novel risk prediction model that can be further replicated and evaluated through application to independent populations.