Establishment and Analysis of a Combined Diagnostic Model of Liver Cancer with Random Forest and Artificial Neural Network

Figure 2

The results of weighted coexpression gene network analysis. An overview of the coexpressed genes in the current study, demonstrating the relevance of gene modules and phenotypes. (a) Screening soft-thresholding powers. (b) Mean network connectivity of soft-thresholding powers used in WGCNA. A soft threshold of 6 is the most suitable value. (c) Cluster dendrogram of the identified coexpression modules. In this figure, each gene is represented as a leaf and corresponds to a color module. Each color indicates that each gene in its corresponding cluster dendrogram belongs to the same module. If some genes have similar changes in expression, then these genes may be functionally related. Moreover, all these genes can further be included into a single module. The gray block represents the genes that do not coexpress with genes of any other color module. (d) Module-trait weighted correlations and corresponding -values for the identified gene module and pathologic type (tumor tissues). The label of color on the right represents the strength of correlation, from 1 (red) to –1 (blue).

(a)

(b)

(c)

(d)