Cancer classification using a novel gene selection approach by means of shuffling based on data clustering with optimization

• An innovative gene selection approach using shuffle method prior to cancer classification is proposed.
• A novel optimization algorithm, COA-GA, is developed by integrating cuckoo optimization algorithm (COA) and GA to enhance classification performance.
• Performance of the COA-GA is analyzed and compared with GA, PSO and COA.
• It is further confirmed that traditional clustering does not have any impact on gene selection and classification performance.
• Optimization based clustering is shown to enhance the accuracy of gene selection and classification.

This research presents an innovative method for cancer identification and type classification using microarray data. The method is based on gene selection with shuffling in association with optimization based unconventional data clustering. A new hybrid optimization algorithm, COA-GA, is developed by synergizing recently invented Cuckoo Optimization Algorithm (COA) with a more traditional genetic algorithm (GA) for data clustering to select the most dominant genes using shuffling. For gene classification, Support Vector Machine (SVM) and Multilayer Perceptron (MLP) artificial neural networks are used. Literature suggests that data clustering using traditional approaches such as K-means, C-means and Hierarchical do not have any impact on classification accuracy. This is also confirmed in this investigation. However, results show that optimization based clustering with shuffling increase the classification accuracy significantly. The proposed algorithm (COA-GA) not only outperforms COA, GA and Particle Swarm optimization (PSO) in achieving better classification performance but also reaches a better global minimum with only few iterations. Higher accuracy is observed to have achieved with SVM classifier compared to MLP in all datasets used.

Figure optionsDownload as PowerPoint slide