Statistical designs for two-color microarray experiments involving technical replication
Computational Statistics & Data Analysis
Robust semiparametric mixing for detecting differentially expressed genes in microarray experiments
Computational Statistics & Data Analysis
| Hi-index | 0.03 |
Statistical microarray data analysis has evolved fast in the last 5 years. However, design issues related to gene expression profiling experiments using two-colour cDNA or oligonucleotide-based microarrays have been investigated only recently. Different approaches and criteria for the choice of efficient designs have been proposed, which apply to a wide range of experiments from simple to complex factorial experimental setups. Here, results from previous research are expanded by showing how contrast vectors and matrices describing the experimental questions under investigation can be used to tailor efficient and economic designs that take into account the estimability of the desired contrasts. The most efficient design among a set of candidate designs is found by applying the e-efficiency criterion from optimal design theory suitable for general factorial designs, which reduces to the comparison of variance factors in the case of vectorial contrasts. These results enable researchers to select efficient designs among several candidate designs for arbitrary complex factorial designs given a set of experimental questions. The corresponding algorithms are implemented in the open source R package daMA available at www.bioconductor.org. The electronic version of this paper can be found at www.sciencedirect.com.