Yeast through the Ages: a Statistical Analysis of Genetic Changes in Aging Yeast

Microarray technology allows for the expression levels of thousands of genes in a cell to be measured simultaneously. The technology provides great potential in the fields of biology and medicine as the analysis of data obtained from microarray experiments gives insight into the roles of specific genes and the associated changes across experimental conditions (e.g., aging, mutation, radiation therapy, drug dosage, ...). The application of statistical tools to microarray data can help make sense of the experiment and thereby advance genetic, biological, and medical research. Likewise, microarrays provide an exciting means through which to explore different statistical techniques. Our paper focuses on the analysis of data from a yeast DNA microarray experiment. The biological question that motivates our research is “What genetic changes in yeast happen over time?” In order to explore the research question of interest we first standardize the data to correct for errors that arise in the data due to biases from the complex microarray procedure. Once we have data that accurately depicts the natural variability in the genes and arrays (as opposed to variability due to technical aspects of the microarray chip), we can focus on our primary interest: the analysis of yeast gene expression to further uncover the quantitative relationship between the gene expression levels and the generation (age) of the yeast cell. We use a statistical tool called predication analysis for microarrays (PAM) [3]; PAM is a classification tool that provides insight into different groupings of the generations of yeast. PAM isolates and identifies specific genes using a threshold value and creates a model to predict the generation for an independent sample array. We also attempt to improve the results obtained from PAM by using t-tests to pre-filter genes before introducing them into the PAM model.