Bic-NSCSA: A hybrid artificial immune system model for DNA microarray bi-clustering

Bi-clustering of microarrays gene expression data is one of the almost young methods which make it possible to cluster genes and samples in microarray datasets simultaneously, and thus their results are more useful. In this paper we propose a hybrid model of clonal and negative selection algorithms to find more accurate bi-clusters in microarrays data. One of the AIS models' shortcomings is slow convergence to global optimum. In proposed method, we use negative selection algorithm as a cooperative method behind of the clonal selection algorithm to overcome the local optimum deficiency and also, add noise-robustness to the proposed model. The algorithm is applied to yeast dataset which is one of the standard typical microarray datasets and compared with several bi-clustering algorithms. The results show significant improvement comparing with the other competing algorithms.