Integrative approaches with applications in eQTL analysis and randomized trials

Purdue University, West Lafayette IN

Investigators

Abstract

Multisource data frequently arise in many real data applications, where different data sources contain complementary information, but each has only limited samples. To make the best use of the limited data from different sources, there is a great need for integrative approaches to jointly analyze all the datasets instead of separately analyzing every single dataset. Motivated by this need, this project will develop new integrative statistical methods to improve parameter estimation accuracy and hypothesis testing power. The results of the project will empower researchers in scientific fields such as genetics, biology, and medicine that face multisource data problems. The project will provide a broad range of interdisciplinary training opportunities for undergraduate and graduate students, especially students from underrepresented groups. The technical goal of this project is to study the integration of multisource data in the following three aspects. First, the principal investigator (PI) plans to build an empirical Bayes regression model for genotype-expression association analysis integrating data from multiple tissues. Second, the project will improve the test of the genotype-expression association via borrowing shared information across genes. Third, the PI will develop a covariate-adjusted method for causal effects on high-dimensional outcomes. The developed methods and results in this project will deepen understanding of genetics and biology, yield a more powerful test of treatment or drug effects on patients, and hence foster significant biological and medical benefits. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.