ÁÖ¼ººÐ ºÐ¼®(Principal Component Analysis, PCA)À» ÀÌ¿ëÇÑ À¯ÀüÀÚ ÁýÇÕ(Gene Set) ±â¹ÝÀÇ ¸¶ÀÌÅ©·Î¾î·¹ÀÌ ÀÚ·á ºÐ¼®
Á¶¼º¹ü1), ±èÁÖÇÑ*1)
1)Seoul National University Biomedical Informatics
Abstract : Gene set-wise analysis provided many researchers a powerful way to understand microarray data. Correlation between gene set members, however, can confuse the gene set-wise testing procedure. In this analysis, the authors performed principal component analysis(PCA) of each gene set to avoid possible disturbing effect of highly correlated gene set members. The resulting orthogonal property of principal components can remedy the high correlation between gene set members. Global test was used for gene set-wise analysis. The analysis of breast cancer data set(GSE 3494 from GeneExpress Ominibus web site) with 387 pathways revealed that 48 pathways showed significance for lymph node metastasis consistently both in raw and transformed data. To measure the degree of correlation between gene set members, we used correlation index(CI). Among 387 pathways, 38 pathways with higher CI than mean CI value showed non-significant result in global test. PCA transformation of original expression matrix of the pathways revealed that these pathways were significant. These result indicated that PCA might be helpful for concerning the correlation structure in the gene set-wise analysis of microarray keyword : gene set-wise microarray data analysis, principal component analysis, correlation index
¿¬Á¦ºÐ·ù
÷ºÎÇÑ ÆÄÀÏ Ã·ºÎÇÑ e-posterÆÄÀÏÅõ°í³»¿ëÀ» »èÁ¦ÇϽ÷Á¸é "»èÁ¦"¹öÆ°À» ´©¸£½Ã°í ´Ù½Ã Á¢¼öÇÏ¿© Áֽñ⠹ٶø´Ï´Ù.
´ëÇÑÀÇ·áÁ¤º¸ÇÐȸ
Copyright (c) 2002 by The Korean Society of Medical Informatics