Innovative Statistical Analysis for Genome-Wide Data with General Interval-Censored Outcomes of Oral Health in Childhood Cancer Survivors

点击翻译按钮获取中文摘要

Project Summary/Abstract This application will study the oral sequelae in childhood cancer survivors from the St. Jude Life cohort and Childhood Cancer Survivor Study cohort. Both disease onset and onset time were collected, but current analyses fail to analyze the disease onset time due to high rate of missing data. DNA samples were collected and sequenced but not analyzed either. We propose innovative ways to analyze the disease onset time in the presence of missing data by considering some onset time as interval-censored, and propose new methods for analyzing interval-censored outcomes with ultrahigh-dimensional genetic covariates. We will perform both single variant-based and rare variant aggregation-based analysis for the whole genome sequencing data. We aim to estimate oral disease dynamics and associated risk factors including environmental factors, genetic factors, and their interaction. Specifically, the aims are: 1). Develop nonparametric and semiparametric screening methods for ultrahigh-dimensional data with interval-censored outcomes; 2). Develop a penalized regression method for data with reduced dimensionality from Aim 1; 3). Apply the methods developed in Aim 1 and Aim 2 to the SJLIFE and CCSS data. We will develop and share multiple user-friendly R codes associated with the new methods. The main objective of the proposed research is to employ the existing methods and develop new statistical procedures to perform appropriate analysis on the whole-genome and oral health data for a deeper understanding of the genetic architecture of tooth development and disease.