December 19, 2014
11:30 AM - 01:00 PM
URMC 2-6408 (K-207 Auditorium)

An Evaluation of Statistical Methods for Differential DNA Methylation Microarray Data Analysis

Dongmei Li, Ph.D.
Clinical and Translational Science Institute

DNA methylation is an intensively studied epigenetic process for elucidating how epigenetic information controls gene expression responsible for a broad range of biological process and diseases. The Illumina HumanMethylation450 BeadChip is becoming a popular platform for quantifying DNA methylation. Statistical methods applicable to Illumina BeadArrays' differential DNA methylation data analysis span a number of approaches such as t-test, rank-based non-parametric test, permutation test, empirical Bayes method, and bump hunting method. Selection of an optimal statistical method, however, can be challenging when multiple methods provide inconsistent results for the same data set. We compare statistical approaches relevant to DNA differential methylation microarray analysis regarding false discovery rate control, statistical power, and stability through simulation studies and real data examples. Results provide guidance for optimal statistical methods selection under different scenarios.

On-Going Research Talk: Computational Ultrasound for Unfocused Acoustic Imaging

Roland Cheng
Department of Electrical and Computer Engineering

A novel acoustic imaging technique, called computational ultrasound imaging, is introduced. It uses randomized, spatio-temporal modulation in lieu of electronic focusing. This simplifies the traditional architecture to allow for cost-effective 2D imaging arrays. Improved (sub-wavelength) image resolution is also demonstrated. Proof-of-concept images are simulated using Field II.