Fig 1.
After the identification of a WGCNA community with a target gene, correlation measures have been applied to build a co-expression gene network. The combined study of the network’s topological properties with information entropy, through a threshold analysis, led to the detection of a pivotal community for the target gene.
Fig 2.
Dendrogram of the network with defined module colors.
This dendrogram was obtained by average linkage hierarchical clustering. The color spectrum underneath the dendrogram indicates the module assignment determined by the Dynamic Tree Cut. The DRD2 gene is contained in the maroon module. Almost 3000 genes are isolated (grey module).
Fig 3.
Adjacency matrix of the maroon module.
This matrix illustrates the similarity between gene expressions using the correlation matrix. Color gradient indicates the strength of the connections between genes.
Fig 4.
Trend of the main topological properties of the network.
The vertical lines highlight an optimal range of threshold values between 0.40 and 0.55 in which three of the four network properties are maximized. For visual clarity, threshold values in which the topological properties are not significant have been omitted.
Fig 5.
Pivotal Module Index, as a function of threshold.
Pivotal Module Index as a function of threshold in the optimal range of the network topological properties obtained using four different community detection methods. Statistical errors were estimated by means of the bootstrap procedure.
Fig 6.
Average Pivotal Module Index (PMI) distribution for the four community detection methods fitted with a linear model. The blue area represents 95% confidence intervals for the fitted function. Statistical errors were estimated by means of the bootstrap procedure. In the top right corner the reduced χ2 and fit parameters are been shown.
Table 1.
Results of the DRD2 community obtained using four community detection algorithms.
The number of genes for the DRD2 community found applying four different community detection algorithms at the selected threshold. The index values next to 1 suggest that the four algorithms are consistent amongst each other and the communities found are similar. Statistical errors were estimated by means of the bootstrap procedure.
Table 2.
Hub genes list of the whole network.
The hub genes of the original module [18]. The first column reports the probe name in BrainCloud. The second column reports the corresponding gene name. The third column contains the strength of hub genes. The fourth column reports the strength of hub genes divided by the maximum value of strength. Colored rows indicate the overlap with the list of genes of detected DRD2 community.
Fig 7.
Betweenness entropy as a function of threshold.
Information entropy based on betweenness as a function of threshold for the network (panel A) and for the DRD2 community (panel B). Both distributions present a maximum threshold value equal to 0.45.
Fig 8.
Dice index for the DRD2 community.
The Dice index as a function of threshold calculated in relation to the DRD2 community obtained with a threshold value equal to 0.45. The statistical errors were estimated by means of the bootstrap procedure.
Table 3.
Results of the strength of DRD2.
Strength of DRD2 per node for the entire network, the DRD2 community selected and for the 1000 randomly selected DRD2 communities.