Generalized Integrated Functional Test for Regional Methylation Rates

TL;DR

The paper introduces GIFT, a new statistical test for detecting regional methylation differences in genomic data, effective even with limited samples and measurement errors.

Contribution

GIFT is a novel functional test that compares group methylation profiles using an ANOVA-like approach, adaptable to various experimental data types.

Findings

GIFT demonstrates good statistical properties in simulations.

GIFT successfully identifies relevant genomic regions in leukemia data.

The method is flexible with different smoothing functions.

Abstract

Motivation: Methods are needed to test pre-defined genomic regions such as promoters for differential methylation in genome-wide association studies, where the number of samples is limited and the data have large amounts of measurement error. Results: We developed a new statistical test, the generalized integrated functional test (GIFT), which tests for regional differences in methylation based on differences in the functional relationship between methylation percent and location of the CpG sites within a region. In this method, subject-specific functional profiles are first estimated, and the average profile within groups is compared between groups using an ANOVA-like test. Simulations and analyses of data obtained from patients with chronic lymphocytic leukemia indicate that GIFT has good statistical properties and is able to identify promising genomic regions. Further, GIFT is likely to work with multiple different types of experiments since different smoothing functions can be used to estimate the functional relationship between methylation percent and CpG site location. Availability and Implementation: Matlab code for GIFT and sample data are available at http://biostat.gru.edu/~dryu/research.html. Contact: rpodolsk@med.wayne.edu or dryu@gru.edu