A Method of Finding Predictor Genes for a Particular Disease Using a Clustering Algorithm

Abstract

Clustering Algorithms are nowadays really important tools in microarray data analysis. The different clustering algorithm generally used in biological science does not take into consideration the underlying probability distribution of the data. In this sense, they are heuristic in nature. In this work we proposed a clustering algorithm based on EM Algorithm. It gives 28% less misclassification than the K-means algorithm (which is mostly use in Bio science). We have also shown on a real data set that this algorithm can be efficiently used for detecting the genes which are responsible for a particular disease.

Keywords:

• EM algorithm
• Microarray
• Model-based clustering

Mathematics Subject Classification:

• 62H30
• 68T10

Acknowledgments

The authors would like to thank the Associate Editor for valuable comments which improved the quality of this article.