원문정보
초록
영어
Microarray gene expression techniques and tools have become of a substantial importance and widely used to analyze the protein-protein interaction (PPI) and gene regulation network (GRN) research in recent years since it can capture the expressions of thousands of genes in a single experiment. Such dataset poses a great challenge for finding association rules in a faster way because of the presence of large number of columns but a small number of rows. Therefore, to meet the challenge of high volume of gene expression and the complexity of microarray data, various data mining methods and applications have been proposed for analyzing gene expressions. However, it is not trivial to extract biologically meaningful information from the huge amount of gene expression data in understanding of gene regulation networks and cellular state, because most cellular processes are regulated by changes in gene expression. Association rule mining techniques are helpful to find relationship between genes, but most of the developed association rule mining algorithms are based on main memory and single processor based techniques which are not capable of handling ever increasing large data and producing result in a faster way. In this paper, we proposed a MapReduce framework for mining association rules from a huge microarray gene expression dataset on Hadoop; which not only overcomes of the main memory bottleneck but also highly scalable in terms of increasing data size. When we apply this new method to the mice lungs and spinal cord microarray compendium data, it identifies a majority of known regulons as well as novel potential target genes of numerous key transcription factors. Extensive experimental results show that our proposed approach is efficient for mining high confident association rules from large microarray gene expression datasets in terms of time and scalability.
목차
1. Introduction and Motivations
2. Problem Statement and Background Study
2.1. Analysis of Microarray Dataset
2.2 Cloud Technologies and the Services for Analyzing Big Dataset
2.3 Apache Hadoop, HDFS and Relevance of MapReduce to Many-Task Computing
2.3 Mining Association Rules
3. Proposed MapReduce Framework
3.1 Proposed Programming Model
3.2 AN EXAMPLE
4. Experimental Results
4.1 Hardware Configurations
4.2 Description of the Datasets
4.3 Performance Analysis
5. Conclusions
Acknowledgement
References