Pooled microarray expression analysis of failing left ventricles reveals extensive cellular-level dysregulation independent of age and sex

Figures

• 

  Fig. 1. Transcriptomes of failing (n = 358) and non-failing (n = 191) left ventricles. (A) Global gene expression landscape summarized by topmost principal components and divided into Principal Groups A and B representing above or below the unbiased decision line. (B) Heat map with unsupervised hierarchical clustering of 3000 genes with greatest inter-sample variance. Each row is a gene while each column is a sample. Heat color represents normalized expression value. Cluster membership was determined at the first tree-node split. Table inset shows the contingency table used for McNemar's test evaluating the concordance of Cluster Membership with global expression Principal Groups. (C) Volcano plot showing differentially expressed genes in failing ventricles. Horizontal dotted line is the Bonferroni p < 0.05 threshold. Vertical dotted lines are the fold-change threshold of 1.50. Genes colored orange and gray are both statistically significant and differ by 1.50 folds. Gray-colored genes do not meet either the fold-change or significance. Table insets are the contingency tables used for two-sided Fisher's exact test determining the relationship between Principal Group membership and…

• 

  Fig. 2. Expression differences between ischemic (n = 159) and dilated (n = 194) cardiomyopathy. (A) Hierarchical clustering of differentially expressed genes, identified by Linear Model for Microarray Data (LIMMA), in heart failure compared to non-failing controls, corresponding to all colored points in Fig. 1C. Table inset is the contingency table for two-sided Fisher's exact test to determine the association between cardiomyopathy subtypes and sub-cluster membership. (B) Volcano plot showing differentially expressed genes in ischemic cardiomyopathy. (C) Differential loci in the ischemic group reaching Bonferroni p < 0.05 threshold.