Statistical Methods for Environmental Genetic Research

University Of Southern California, Los Angeles CA

Investigators

Linked publications & trials

Abstract

Over the past 13 years the Southern California Children's Health Study (CHS), one of the few long term[unreadable] studies with focus on children's health, has demonstrated that children's respiratory health is adversely[unreadable] affected by long term exposures to air pollution. It has also demonstrated the important role of genetics in[unreadable] modulating the impact of the environment. These results have had significant policy implications for[unreadable] protecting children's health, because they have added to our prior knowledge base, which had mainly[unreadable] focused on adults. These results were made possible through the use of novel and innovative multi-level[unreadable] modeling techniques, including many that were developed by CHS statisticians to address the various levels[unreadable] of comparisons and challenging recurring issues that were necessitated by the scientific questions under[unreadable] study. In this project, we continue to develop new statistical methods that are motivated by the general[unreadable] theme of this renewal application towards integrated analysis of the exposure, lung function, asthma, genetic[unreadable] and biomarker data. We propose to develop new statistical techniques in four interrelated areas. These[unreadable] include (i) extension of our work on exposure modeling and measurement error, with renewed focus on[unreadable] exploiting spatial correlations for intra-community variation and joint multivariate modeling of several[unreadable] pollutants, (ii) development of techniques for analyzing the high volume of genetic data that will be generated[unreadable] by Project 2, including methods for analyzing multiple single nucleotide polymorphisms (SNPs) per candidate[unreadable] locus and multiple candidate loci per pathway, (iii) development of new latent variable based flexible multistate[unreadable] modeling techniques that can handle joint analysis of multiple outcomes including continuous lung[unreadable] function data and time-to-event asthma diagnosis and related phenotypes, giving us a tool to better[unreadable] understand factors that potentially affect a spectrum of respiratory health outcomes in children in ways that[unreadable] account for possible outcome misclassification, and (iv) development of novel techniques for integrated[unreadable] analysis of the health outcome, genetic, biomarker and exposure data, where the integration of the scientific[unreadable] evidence about the chronic effects of environmental exposure on children's health is examined by drawing[unreadable] information from Projects 1 and 2. Although the topics we pursue will be motivated by CHS data and issues,[unreadable] the methods we develop will also be applicable to a broad range of epidemiological studies.