Mean, median and tri-mean based statistical detection methods for differential gene expression in microarray data

The detection of differential gene expression in microarray data can recognize genes with significant alteration of expression level with regard to varying experimental environment. Traditional differential gene expression detecting methods work on the assumption that all cancer samples are over-expressed compared with normal samples and need to define the key criterion with the mean of sample data. In recent proposed methods, one often considers the situation that only a subgroup of cancer samples are over-expressed and only the key criterion with the median and median absolute deviation is required. We proposed a detecting method for over-expressed cancer subgroup by defining the key criterion with tri-mean and tri-mad. Numerical experiments on public microarray data indicate that the improved method outperforms the compared methods.