Network nodes (voxels, vertices, etc.) have roles in the network. For example, Guimera & Amaral suggest seven universal roles. Role 1: nodes only connected to others in their own community ( ultra-peripheral nodes ). Role 2: nodes connected mostly only to others in their own community ( peripheral nodes ). Role 3: nodes connected to many other communities ( non-hub connector nodes ). Role 4: nodes with homogenously distributed connections to all communities. Nodes with a high within-module degree classify as hub nodes. Role 5: Some hub nodes are only really connected to others in their own community ( provincial hubs ). Role 6: Other hub nodes are also highly connected to other communities ( connector hubs ). Role 7: Hub nodes with homogeneously distributed links among communities are kinless hubs.

Within-module degree

Using this R code. And I call the R code from the command with RInvoke.. The command line arguments are (1) the participant ("ss") name, (2) the Condition number, and (3) the tree number. The script will call a file that contains the community identifiers (see line 14).

Participation coefficient

Using this R code. And I call the R code from the command with RInvoke. Structured the same as above. The command line arguments taken by the R code (in order) it needs are (1) the participant ("ss") name, (2) the Condition number, and (3) the tree number. A sample command line:

./RInvoke.sh 16.partcoef.R CLFR 1 4

Get Node Roles

After determining the within-module degrees and participation coefficients, each voxel can be identified with a role. I use the conventions described above. This script iterates through the within-module degree and participation coefficient outputs and assigns the role for every voxel.

Relevant papers

• Functional cartography of complex metabolic networks