Gene Expression Data Analyse Sofeware Based On Multi-objective Artificial Bee Colony Biclustering

Due to the advent of High-throughput microarray technology, it generated large amountsof gene expression datasets. To discover novel and useful knowledge from those datasetsmore effectively and efficiently, there is an increasing need to develop more powerful datamining algorithms and relevant softwares. Biclustering, which clusters both microarraygenes and conditions simultaneously, can detect the submatrices which their genes heavilycorrelates with their conditions. The submatrices are called biclusters. While doing theresearch, we found that, To detect higher quality biclusters, researchers always have tooptimize several conflicting objectives at the same time. However, few softwares can offercomputing service of biclustering algorithms based on muti-objective optimization, and fewsource codes were released on the internet. Besides, the data processing in the past isrelatively independent, unautomatically and unresuablily. Hence, In this paper, we designedand realized Multi-objective Artificial Bee Colony Biclustering. What is more, we designedand implemented a gene expression data analyse sofeware based on the algorithm.The main research work we have done can be summarized as follows:(1) we present aframework which applies Multi-objective Artificial Bee Colony Algorithm to detectebiclusters. Several significant designs in terms of coding techniques、different search strategyfor bees with different labor、measurement of the overall quality of the foods、strategy fortruncating the external archive were made to enchance the local and global search ability ofthe algorithm for better foods, and to ensure the biclusters can get colser to global optimumsolutions. To evaluate the performance of the proposed algorithm, we respectivelyimplemented the propesed algorithm on the yeast and human B-cells datasets, and comparedthe results with several biclustering algorithms.Results showed that the quality of biclustersfound by the propesed algorithm are better than others. And the propesed algorithm have goodpopulation diversity and rapid convergence.(2) we design and implement a gene expressiondata analyse sofeware based on the propesed algorithm. It integrates the computing functionof biclustering algorithm based on muti-objective optimization, biclusters managementfunction and biclusters analysis function. Implement results showed that the sofwareimproved the work efficiency greatly, and achived the expected goals.