Computing a Predictor Set Influence Zone through a Multi-Layer Genetic Network to Explore the Role of Estrogen in Breast Cancer

Modeling inter-relationships of genes over a specific genetic network is one of the most challenging studies in systems biology. Among the families of models proposed one commonly used is the discrete stochastic, based on conditionally independent Markov chains. In practice, this model is estimated from time sequential sampling, usually obtained by microarray experiments. In order to improve the accuracy of the estimation method, we can use biological knowledge. In this paper, we decided to apply this idea to study the role of estrogen in breast cancer proliferation. The n-influence zone of a set S of genes in a given multi-layer genetic network is a set L of genes regulated, directly or indirectly, by genes in S, after at most n-1 layers. In this manuscript we describe a new approach for computing the n-influence zone of S through the estimation of a multi-layer genetic network from gene expression time series, measured by microarrays, and biological knowledge. Using seed genes related to cell proliferation, our method was able to add to the third layer of the network other genes related to this biological function and validated in the literature. Using a set of genes directly influenced by estrogen, we could find a new role for cell adhesion genes estrogen dependent. Our pipeline is user-friendly and does not have high system requirements. We believe this paper could contribute to improve the data mining for biologists in microarray time series.