This paper discusses different approaches for integrating biological knowledge in gene expression analysis. Indeed we are interested in the fifth step of microarray analysis procedure which focuses on knowledge discovery via interpretation of the microarray results. We present a state of the art of methods for processing this step and we propose a classification in three facets: prior or knowledge-based, standard or expressionbased and co-clustering. First we discuss briefly the purpose and usefulness of our classification. Then, following sections give an insight into each facet. We summarize each section with a comparison between remarkable approaches.