A simple analysis of gene expression and variability in gene arrays based on repeated observations

At the present time there is an explosion of research in the area of gene arrays, and their application for detection of genes related to disease as well as its therapeutic manipulation. However, as individual arrays are expensive, comparisons of gene expression are often not repeated. In the current study, gene array experiments were repeated multiple times in order to understand the variability associated with measurements of gene expression. By focusing upon the pharmacologically important target of prostate cancer cell detachment, the current study employed multiple repeats of gene array experiments. This was used as a model system to demonstrate the utility of the experimental approach and statistical methods used.

To identify genes involved in detachment of prostate cancer cells (a prerequisite for metastases), we analyzed gene expression changes in metastatic variant PC3-M cells undergoing spontaneous detachment in culture. The data were interpreted using an elementary statistical approach. The between-experiment and within-repeated-observations variability in expression of 3582 genes possibly related to prostate cancer was also evaluated.

One important gene related to prostate cell detachment was identified, based on the magnitude of its change in expression, as measured by a ratio of the expression after cell detachment and expression before detachment. On average, the variation between experiments was greater by about 30 to 40% than the variation between repeated observations.

These findings have implications relating to the use of gene arrays to detect variance of gene expression, and should be taken into consideration in the prospective design of array experiments.