An integrated time series gene expression data analysis pipeline with a fuzzy clustering method to assess expression patterns

Technical improvements in high-throughput gene expression experiments are making possible to obtain high quality time series whole-genome expression data sets. This valuable source of information may describe the unfolding biological processes during the development stages, the cell cycle or the immune response of an organism. In order to fully explore this type of data we developed an integrated time series gene expression analysis pipeline. The resulting method detects differentially expressed genes, cluster co-expressed genes, unveil hidden gene expression patterns, identify over represented biological function categories and infer gene regulatory networks. Some of the methods integrated in our pipeline are an empirical Bayes model, a noise robust fuzzy clustering and graphical Gaussian model. The use of this pipeline to analyze the human adenovirus infection process allowed us to discover new insights and hypothesis. No previous exhausted explorations including features as fuzzy clustering or regulatory network inference have been used on this biological phenomena data before.