AIC'05 Proceedings of the 5th WSEAS International Conference on Applied Informatics and Communications
A new maximum-relevance criterion for significant gene selection
PRIB'06 Proceedings of the 2006 international conference on Pattern Recognition in Bioinformatics
A multi-information based gene scoring method for analysis of gene expression data
Transactions on Computational Systems Biology V
| Hi-index | 0.00 |
Microarray technology is a recent development inexperimental molecular biology which can producequantitative expression measurements for thousands ofgenes in a single, cellular mRNA sample. These manygene expression measurements form a composite profileof the sample, which can be used to differentiate samplesfrom different classes such as tissue types or treatments.However, for the gene expression profile data obtained ina specific comparison, most likely only some of the geneswill be differentially expressed between the classes, whilemany other genes have similar expression levels.Selecting a list of informative differential genes fromthese data is important for microarray data analysis. Inthis paper, we describe a framework for selectinginformative genes, called Ranking and Combinationanalysis (RAC), which combines various existinginformative gene selection methods. We conductedexperiments using three data sets and six existing featureselection methods. The results show that the RACframework is a robust and efficient approach to identifyinformative gene for microarray data. The combinationapproach on two selecting methods almost alwaysperformed better than the less efficient individual, and inmany cases, better than both. More significantly, whenconsidering all three data sets together, the combinationapproach, on average, outperforms each individualfeature selection method. All of these indicate that RCAmight be a viable and feasible approach for themicroarray gene expression analysis.