Troubleshooting methods for microarray gene expression analysis in the onset of diabetic kidney disease

Abstract

Introduction: Microarrays have become the standard technique for discovering new genes involved in the development of (kidney) disease. Diabetic nephropathy is a frequent complication of diabetes and is characterized by renal fibrosis. As the pathways leading to fibrosis are initiated early in diabetes and in the current study, we aimed at identifying genes associated with renal fibrosis in the first week after induction of diabetes in the rat streptozotocin (STZ) model. Methods: Conventional microarray analysis methods comparing gene expression to a common reference are not very suitable for time series as gene lists for all time point are very heterogeneous. We therefore sought an analysis technique that would allow us to easily find genes that we either substantially up or down regulated during the first week of diabetes. In the new method, the normalized expression of individual genes was plotted in time. Subsequently, the area under the curve (AUC) was calculated to quantify the overall level of changes in expression of individual genes. Results: AUCs for all genes were plotted in a histogram showing a normal distribution with a mean of close to 0, indicating no change in expression for the majority of genes. Genes with AUCs outside 3 standard deviations of the mean were considered significantly different from control. Discussion: Using this technique, a total of 290 genes were found to be significantly changed in the first week of diabetes. Data on a subset of genes were confirmed by real-time PCR, indicating the validity of the employed new analysis method.