AbstractWe analyze DNA microarray time series describing gene expression in unicellular organisms subject to external perturbations or along developmental stages of higher eukaryotes. Using a translation-invariant and scale-invariant distance measure to compare the gene expression profiles, we show that peaks in the average distances are noticeable and are localized around specific time points. These coincide with the transition between developmental phases or follow external perturbations. This approach can be used to identify automatically, from microarray time series alone, the presence of such perturbations or transitions in arbitrary cell systems. Our results reveal the striking similarity between the gene expression responses to these very different phenomena. We set up a clustering method that uses the abovementioned distance to classify the genes within each stage and applied it to the development of the Drosophila embryo. The evolution of the average cluster expression levels was analyzed using coupled linear and non-linear differential equations. Different model structures and schemes for parameter identification and reduction were tested. The models obtained were compared on the basis of their abilities to reproduce the data, to keep realistic gene expression levels when extrapolated in time, to show the biologically expected robustness with respect to parameter variations, and to contain as few parameters as possible. A family of non-linear models reached all the objectives. It defined networks with an average of two connections per node.