Fig. 1

Identifying critical lncRNA in ovarian cancer EMT. a Ovarian cancer patients (n = 320) with genomic and molecular profiling data that classified into epithelial (Epi; n = 231) or mesenchymal (Mes; n = 89) subtypes were selected for analysis. b Heatmap of 386 genes that were differentially expressed in the mesenchymal subtype compared with the epithelial subtype. c Inferring deregulatory programs from ovarian cancer profiling data. Change in mRNA expression is modeled as linear function of the gene’s DNA methylation, copy number, and lncRNA expression. d, e Systematic prediction of EMT-linked lncRNA from the lncRNA-gene association information obtained from the linear model. d The lncRNA that had significantly enriched association with the differentially expressed genes (n = 25, red dots; top 5 lncRNA labeled) were inferred as EMT associated. Remaining lncRNA were represented by gray dots. The X-axis with four different colors represent major annotation classes of the selected lncRNA (n = 120). The Y-axis denotes which lncRNA had enriched association with the differentially expressed genes compared with non-differentially expressed genes. e Filtering of high confidence EMT-linked lncRNA (n = 4; blue dots with labels) based on their aberrant expression (X and Y-axis) in EMT and conservation score (Z-axis). Gray dots represent remaining lncRNA. f Heatmap shows significantly enriched association of the inferred lncRNA with EMT-linked pathways. For d and f, P-values determined by BH adjusted hypergeometric test