Predicting the Functions of Long Noncoding RNAs Using RNA-Seq Based on Bayesian Network

Figure 7

(a) Regulatory networks between lncRNAs and protein-coding genes were, respectively, constructed using RNA-seq data of breast cancer, prostate cancer, and psychiatric disorders based on the Bayesian network method. Red nodes represent protein-coding genes, green nodes represent lncRNAs, and edges represent regulatory relationships. (b) Heatmap representing the GO terms of lncRNAs in four data sets. For each lncRNA, detected GO terms are indicated in red. I denotes psychiatric disorders, II denotes breast cancer, III denotes prostate cancer (GSE22260), and IV denotes prostate cancer (GSE25183). (c) Heatmap shows the shared functions of lncRNAs in four data sets (columns). Color key represents the number of data sets sharing the same GO terms for each lncRNA (rows). (d) Functional heatmap (left panel) representing the GO terms (rows) of ENST00000448587 in four data sets (columns). Detected GO terms in each data set are indicated in red. Venn diagram (right panel) showing the number of overlapped GO terms of ENST00000448587 in four data sets. (e) Functional heatmap (left panel) representing the GO terms (rows) of ENST00000411553 in four data sets (columns). Detected GO terms in each data set are indicated in red. Venn diagram (right panel) showing the number of overlapped GO terms of ENST00000411553 in four data sets.

(a)

(b)

(c)

(d)

(e)