Biclustering is a thriving and of paramount task in many biomedical applications. Indeed, the biclusters aim, among-others, the discovery of unveiling principles of cellular organizations and functions, to cite but a few. In this paper, we introduce a new algorithm called, BiARM, that aims to efficiently extract the most meaningful, low overlapping biclusters. The main originality of our algorithm stands in the fact that it relies on the extraction of generic association rules. The reduced set of association rules faithfully mimics relationships between sets of genes, proteins, or other cell members and gives important information for the analysis of diseases. The effectiveness of our method has been proved through extensive carried out experiments on real-life DNA microarray data.