Network analysis metrics and algorithms have been designed to identify network hub nodes and modules in a scale-free network.
The biological networks are scale-free but converge in regulatory nodes and modules, such as major transcription factors and receptors [ 13, 14].
To investigate the prognostic potential of critical nodes and modules, we examined the ability of their corresponding genes to distinguish between poor and good outcomes after MI (specified below).
For each combination of drug and co-expression module, relevance scores were calculated for each node and edge in the corresponding response network, representing the expected number of times the node or edge is visited by random walks between the drug node and module nodes.
This process yielded 70 modules with 3 nodes, 35 modules with 4 nodes, 18 modules with 5 nodes, 12 modules with 6 nodes, and 56 modules with 7 or more nodes.
According to their location in the modules (when k = 3), these hubs can be divided into two categories: module hubs (192 nodes) and inter-module hubs (69 nodes).
To get additional nodes we used Genbank (adding nodes from literature and checking their sequence), STRING (predict edges, i.e. interactions between nodes) and network module analysis [ 14] to expand the network.
Because the codelength only depends on the rates of node visits and module transitions, the map equation framework is agnostic to the origin of the flow.
We first sought high traffic nodes and network modules showing strong differential expression capability, at the gene set level, across good and poor clinical outcome.
We anticipate that our work provides new insights to the structure of biological functional networks and highlights the aspects that need to be considered in the prediction of regulatory nodes and functional modules from a multilevel and heterogeneous network context.
