I used WGCNA to find the associated gene modules with the different subtypes of a given cancer as trait. I obtained multiple modules associated with some of the cancer subtypes as shown in . Next, I plotted PCA to visualize how different subtypes can be separated by the genes of green (about 450 genes) and turquoise (about 1000 genes) modules. Based on the PCA, the green module can separate the subtype D from other subtypes, well. However, the PCA plot using the turquoise module genes is a bit weird as you can see . Given the relatively high positive and negative correlation of the turquoise module with subtypes A and D, respectively, I expected that the PCA plot can clearly separate these subtypes. Please share with me any of your thoughts and suggestions about this PCA.