3.6 Correlation
Analysis for Each Node
in the Module
The information of each node is extracted by dividing the column
Cross-references into multiple columns using the character “|”.
To keep the information of gene symbols and Entrez IDs, we
rename the column to NodeName and ENTREZ, respectively.
And we save the file as nodes.csv. We merge the nodes.csv and
annot.csv to find the corresponding probe IDs of each module
node (Fig.13a, b). We also merge them to the normalized expres-
sion data eDat.csv to call the individual expression data of a module
node (Fig.13c). We save the file and prepare it as corNod.csv for
correlation analysis with the columns of ENTREZ and expression
values (Fig.13d). We calculate the correlation of module nodes by
psychpackage according to their expression data (Fig. 14)(see
Note 3). The node pairs with correlation coefficient values more
than 0.9 and adjustedPvalues less than 0.01 are chosen for further
analysis.
Finally, a submodule of six nodes with EDD protein as the cen-
ter tested by the correlation analysis is manually extracted (Table1).
The six proteins represented by the submodule of 6 nodes with
significant correlations will be selected as targets for subsequent
drug design. Their 3D protein structures (if available) can be
collected from the Protein Data Bank.Fig. 11Network visualization using Cytoscape
192 Sze Chung Yuen et al.