Identification of Co-regulated Signature Genes in Pancreas Cancer- A Data Mining Approach
ICIC '08 Proceedings of the 4th international conference on Intelligent Computing: Advanced Intelligent Computing Theories and Applications - with Aspects of Theoretical and Methodological Issues
GUEST EDITORIAL: Computational intelligence in solving bioinformatics problems
Artificial Intelligence in Medicine
| Hi-index | 0.00 |
The main techniques currently employed in analyzingmicroarray expression data are clustering andclassification. In this paper we propose to use associationrules to mine the association relationships amongdifferent genes under the same experimental condition.These kinds of relations may also exist across manydifferent experiments with various experimental conditions.In this paper, a new approach, called LIS-growth (LargeItemSet growth) tree, is proposed for mining themicroarray data. Our approach uses a new data structure,JG-tree (Jiang, Gruenwald), and a new data partitionformat for gene expression level data. Each data valuecan be presented by a sign bit, fraction bits and exponentbits. Each bit at the same position can be organized into aJG-tree. A JG-tree is a lossless and compression tree. Itcan be built on fly, a kind of real-time compression for bitsstring. Based on these two new data structures it ispossible to mine the association rules efficiently andquickly from the gene expression database. Our algorithmwas tested using the real-life datasets from the geneexpression database at Stanford University.