Gene expression: an example of Statistical analysis of real-time PCR data

Real-time Polymerase Chain Reaction (RT-PCR) is one of the most sensitive and reliably quantitative methods for measuring gene expression levels. Despite this technique is broadly applied in biomedical sciences, data processing and statistical procedures for the analysis of RT-PCR data still present some lacks in reliability.In our work we analyze a data set from a genetic medicine experiment whose aim is to investigate the expression level of 225 genes in 2 types of cells which are possible candidates for tumour acceleration. In this experiment 2 biological replicates for 15 mice were considered, using 3 different experimental platforms.Our goal is to estimate a model which accounts for the complex design of a RT-PCR experiment, thus including sample effect, gene effect, treatment effect, card effect and interactions. The main interest is the interaction between gene and sample.With respect to standard techniques our model allows for a better estimation of the variance with less probability of non significant result in case of true differences.