Fig 1.
An overview of the study.
Table 1.
Datasets used in the study.
Fig 2.
Network topology analysis for various soft thresholds.
The left panel shows the scale-free fit index (y-axis) as a function of the soft threshold value (x-axis); the right panel shows the average connectivity (y-axis) as a function of the soft threshold value (x-axis).
Fig 3.
Correlation matrix resulting from WGCNA applied on the training dataset.
Here, each cell indicates the Pearson correlation and p-value resulting from the association between the respective module eigengenes (row) and phenotype (column).
Table 2.
Selected significant modules and the number of genes obtained on the dataset.
Table 3.
Performance comparison of clustering algorithms.
Table 4.
Summary of significant submodules detected by the best performing FN and Spectral clustering algorithms.
Fig 4.
The gene-term graph in which the up / down regulated genes of the terms that are significant terms (A. KEGG pathway, B. Cancer Hallmark term, and C. GO-BP) for the 3rd module of the FN algorithm.
Fig 5.
The gene-term graph in which the up / down regulated genes of the terms that are significant terms (A. KEGG pathway, B. Cancer Hallmark term, and C. GO-BP) for the 1st module of the Spectral algorithm.
Table 5.
The results of drug screening that targets biomarker proteins identified a result of clustering analysis.
The expression type column shows gene expression change of the gene. The action type was chosen according to the expression status of a target.