Statistical Tools for Whole-Genome Prediction of Complex Traits and Diseases

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Genome Wide Association Studies (GWAS) have uncovered an unprecedented number of variants associated with important health-related traits and diseases. Evidence from these studies suggests that most clinically relevant traits have complex genetic architectures. Whole Genome Prediction (WGP) is a predictive approach, primarily developed and tested in the field of animal breeding, designed to confront some of the challenges emerging in the prediction of complex traits and diseases. Implementing WGP requires specialized software, which is not available in standard statistical packages. In our research projects involving plant, animal and more recently human data, we have developed, tested and used statistical software for parametric and semi-parametric WGP. In this project we propose to integrate and further develop this software in ways that will improve its value for applications with human data. We will integrate parametric and semi-parametric procedures for WGP into a unified framework and will deliver software that could be used with un-censored, censored, binary and ordinal traits. The software produced in this project will be delivered as an R-package and will be integrated into GenePattern; a bioinformatics platform where users will be able to develop analysis pipelines by combining our software with other bioinformatics tools. PUBLIC HEALTH RELEVANCE: Genome Wide Association Studies (GWAS) have uncovered an unprecedented number of variants associated with important health-related traits and diseases. Evidence from these studies suggests that most clinically relevant traits have complex genetic architectures. Whole Genome Prediction (WGP) is a predictive approach, primarily developed and tested in the field of animal breeding, designed to confront some of the challenges emerging in the prediction of complex traits and diseases. We believe that this methodology offers great opportunities to advance our ability to predict genetic predisposition to complex human traits and diseases. Implementing WGP methods requires specialized software, which is not available in standard statistical packages. In our research we have developed, tested, and used statistical software for parametric and non-parametric WGP. The proposed project will integrate these software into a unified framework, will further develop these packages by implementing additional regression methods, and will extend the software to handle traits often encountered in human applications such as censored, binary and ordinal outcomes. The software developed in this project will be integrated into R and into GenePattern, a bioinformatics workflow platform which will enable users to integrate our software with other bioinformatics tools.