Identifying Differentially Expressed genes in the Absence of Replication

In microarray data analysis, the comparison of gene expression levels in different conditions and selection of biologically relevant genes are essential tasks. In this study, we propose a novel statistical procedure based on standardised conditional residuals from a linear mixed-effects model which allows comparison of conditions, even if only one replicate per experimental condition is available. We illustrate this method by using three publicly available datasets. We show that this method can be extended to handle more complex designs. Finally, simulations show that the tests developed have good statistical power to detect true differences among conditions at the gene level.

Publication year:2013