©1996-2009 All Rights Reserved.
Online Journal of Bioinformatics
OJBTM
Online Journal of Bioinformatics©
Volume
10 (1):67-73, 2009
Multiple contrast test for
detecting monotonic dose-response relationship and FDR-adjusted confidence intervals
for selected parameters in a microarray Setting
Lin D1, Shkedy Z1, Burzykowski
T1, Yekutieli D2, De Bondt A3, Göhlmann WHH3, Talloen W3, Bijnens
L3
1- Hasselt University, I-BioStat, Universitaire Campus,
Building D, B 3590 Diepenbeek, Belgium 2- Department
of Statistics and Operation Research, School of Mathematical Sciences, Tel Aviv
University, Ramat Aviv, Tel Aviv, 69978, Israel, 3-
J&JPRD - Biometrics and Clinical Informatics, Beerse,
Belgium
SUMMARY
Lin D, Shkedy
Z, Burzykowski T, Yekutieli
D, De Bondt A, Goehlmann H,
Talloen W, Bijnens L.,
Multiple contrast test for detecting monotonic dose-response relationship and
FDR-adjusted confidence intervals for selected parameters in a microarray Setting, Online J Bioinformatics, 10(1):67-73,2009.
Dose-response microarray experiments consist of
monitoring expression levels of thousands of genes with respect to increasing
doses of the compound treatment under investigation. In this paper we discuss a
microarray dose-response experiment in which gene
expression data are available for a control and several treatment doses. That fact that the gene expression increases/decreases with the
increasing doses constitute the active dose-response relationship in this
setting. We aim at comparing the (relative) mean difference in gene
expression between higher doses and the control. Especially, we direct this
test by using Marcus' multiple contrasts to obtain the isotonic means, as
proposed byBretz (2006). Moreover, we show an
application of the multiple ratio tests, discussed by Dilba
et al. (2005), to the data. Furthermore, we construct simultaneous
confidence intervals for a selected subset of genes following the ratio tests, Benjamini and Yekutieli (2005)
addressed the issue of multiplicity due to effect of testing and selecting the
parameters of interest. The Benjamini and Hochberg
(1995) procedure for controlling FDR is applied to address the multiple testing
issue. To construct confidence intervals for selected parameters, the False
Discovery Rate (FDR) adjusted procedure (Benjamini
and Yekutieli, 2005) is applied. The case study used
for illustration is a dose-response microarray
experiment with 12 samples (three arrays at each of four dose levels) and
arrays consisting of 16998 genes.
Keywords: Microarray; Dose Response; Ratio test; False Discovery Rate (FDR) Adjusted Multiple Confidence Intervals (CI); Selected Parameters.