RNA-seq datasets from B cells, H1 ESCs, and brain (see Table S1) were clustered by lincRNA expression levels. LincRNAs with FPKM>10 in one or more of the 7 RNA-seq datasets analyzed in Figure 3B were used to generate the heatmap and dendrogram. These 7 datasets were chosen for this analysis because they have replicates from each tissue and have deep read counts for all replicates (Table S1), important features for accurate measurement of differential expression. Using Gene Cluster 3.0, FPKM values were log2 transformed and the genes (rows) and samples (columns) were normalized by multiplying each log2 transformed FPKM value by a scale factor such that the sum of the squares of the values in each row and column are 1.0. Euclidean distance using centroid linkage was calculated for all samples and the heatmap and dendrogram was generated with Java TreeView. Red corresponds to fully induced expression and blue corresponds to fully repressed expression.
