Home // BIOTECHNO 2011, The Third International Conference on Bioinformatics, Biocomputational Systems and Biotechnologies // View article

UMPIRE: Ultimate Microarray Prediction, Inference, and Reality Engine

Authors:
Jiexin Zhang
Kevin Coombes

Keywords: gene expression; microarray; simulation; class prediction; multi-hit theory of cancer

Abstract:
High-throughput measurements of gene expression pose a challenge to analysts attempting to learn models that predict treatment response or survival. One possible explanation for the lack of significant progress in this area is the limited sample size of most experiments. Realistic simulations could help with the development and assessment of analytical methods; however, existing simulation tools have focused more on the technology and less on the biological complexity. In this paper, we introduce a package of simulation tools to address this problem. Our model incorporates additive and multiplicative noise, transcriptional activity or inactivity, and block correlation structures. More importantly, it models the multi-hit theory of cancer via latent variables that link gene expression, binary outcome, and survival data. We illustrate the use of the simulation package by showing that standard analysis methods (i.e., univariate Cox models) are only likely to recover the true structure with more samples than are included in most current studies of survival.

Pages: 121 to 125

Copyright: Copyright (c) IARIA, 2011

Publication date: May 22, 2011

Published in: conference

ISSN: 2308-4383

ISBN: 978-1-61208-137-3

Location: Venice/Mestre, Italy

Dates: from May 22, 2011 to May 27, 2011