The systems biological method WGCNA was applied to the gene co-expression network of TCGA-UM (23 (link)). The following outlines the steps are taken: exclusion of genes with missing values using the “goodSamplesGenes” function, grouping of tumor samples, deletion of outliers, and establishment of a cut line of 100. The optimal soft threshold for adjacency calculation was determined using graphical methods. An adjacency matrix was generated from the expression matrix to determine the genetic interconnectedness of the network. The topological overlap matrix (TOM) was then constructed from the adjacency matrix. Hierarchical clustering was performed using an average linkage approach and the differences in TOM. The hierarchical clustering tree was dynamically pruned to identify similar modules with high correlation coefficients (r > 0.25). Pearson’s correlation test was applied to examine the relationship between eigengenes and clinical characteristics. The modules containing genes with the most significant correlations to clinical traits, such as glycosylation score, survival status, and survival time were selected for further investigation.
Free full text: Click here
