community.tree {bio3d}

Reconstruction of the Girvan-Newman Community Tree for a CNA Class Object.

Description

This function reconstructs the community tree of the community clustering analysis performed by the ‘cna’ function. It allows the user to explore different network community partitions.

Usage

  community.tree(x, rescale=FALSE)

Arguments

x

A protein network graph object as obtained from the ‘cna’ function.

rescale

Logical, indicating whether to rescale the community names starting from 1. If FALSE, the community names will start from N+1, where N is the number of nodes.

Details

The input of this function should be a ‘cna’ class object containing ‘network’ and ‘communities’ attributes.

This function reconstructs the community residue memberships for each modularity value. The purpose is to facilitate inspection of alternate community partitioning points, which in practice often corresponds to a value close to the maximum of the modularity, but not the maximum value itself.

Value

Returns a list object that includes the following components:

modularity

A numeric vector containing the modularity values.

tree

A numeric matrix containing in each row the community residue memberships corresponding to a modularity value. The rows are ordered according to the ‘modularity’ object.

num.of.comms

A numeric vector containing the number of communities per modularity value. The vector elements are ordered according to the ‘modularity’ object.

Author(s)

Guido Scarabelli

See Also

cna, network.amendment, summary.cna

Examples

# PDB server connection required - testing excluded

if (!requireNamespace("igraph", quietly = TRUE)) {
   message('Need igraph installed to run this example')
} else {

###-- Build a CNA object
pdb <- read.pdb("4Q21")
modes <- nma(pdb)
cij <- dccm(modes)
net <- cna(cij, cutoff.cij=0.2)


##-- Reconstruct the community membership vector for each clustering step.
tree <- community.tree(net, rescale=TRUE)

## Plot modularity vs number of communities
plot( tree$num.of.comms, tree$modularity )

## Inspect the maximum modularity value partitioning
max.mod.ind <- which.max(tree$modularity)

## Number of communities (k) at max modularity
tree$num.of.comms[ max.mod.ind ]

## Membership vector at this partition point 
tree$tree[max.mod.ind,]

# Should be the same as that contained in the original CNA network object
net$communities$membership == tree$tree[max.mod.ind,]

# Inspect a new membership partitioning (at k=7)
memb.k7 <- tree$tree[ tree$num.of.comms == 7, ]

## Produce a new k=7 community network  
net.7 <- network.amendment(net, memb.k7)
plot(net.7, pdb)
#view.cna(net.7, trim.pdb(pdb, atom.select(pdb,"calpha")), launch=TRUE )

}