Interpreting microarray data to build models of microbial genetic regulation networks

Microarrays and DNA chips are an efficient, high-throughput technology for measuring temporal changes in the expression of message RNA (mRNA) from thousands of genes (often the entire genome of an organism) in a single experiment. A crucial drawback of microarray experiments is that results are inherently qualitative: data are generally neither quantitatively repeatable, nor may microarray spot intensities be calibrated to in vivo mRNA concentrations. Nevertheless, microarrays represent by the far the cheapest and fastest way to obtain information about a cell's global genetic regulatory networks. Besides poor signal characteristics, the massive number of data produced by microarray experiments pose challenges for visualization, interpretation and model building. Towards initial model development, we have developed a Java tool for visualizing the spatial organization of gene expression in bacteria. We are also developing an approach to inferring and testing qualitative fuzzy logic models of gene regulation using microarray data. Because we are developing and testing qualitative hypotheses that do not require quantitative precision, our statistical evaluation of experimental data is limited to checking for validity and consistency. Our goals are to maximize the impact of inexpensive microarray technology, bearing in mind that biological models and hypotheses are typically qualitative.