A Top-K formal concepts-based algorithm for mining positive and negative correlation biclusters of DNA microarray data

Analyzing and understanding large and complex volumes of biological data is a challenging task as data becomes more widely available. In biomedical research, gene expression data are among the most commonly used biological data. Formal concept analysis frequently identifies deferentially expressed genes in microarray data. Top-K formal concepts are effective at producing effective Formal Concepts. To our knowledge, no existing algorithm can complete the difficult task of identifying only important biclusters. For this purpose, a new Top-K formal concepts-based algorithm for mining biclusters from gene expression data is proposed: Top-BicMiner. It extracts biclusters’ sets with positively and negatively correlated genes according to distinct correlation measures. The proposed method is applied to both synthetic and real-life microarray datasets. The experimental results highlight the Top-BicMiner’s ability to identify statistically and biologically significant biclusters.