meta.communities: Produce a Set of Meta-Communities

In alextkalinka/linkcomm: Tools for Generating, Visualizing, and Analysing Link Communities in Networks

Description

This function returns meta-communities by hierarchically clustering link communities based on their number of shared nodes.

Usage

meta.communities(x, hcmethod = "ward.D", deepSplit = FALSE)

Arguments

x	An object of class linkcomm or OCG.
hcmethod	A character string naming the hierarchical clustering method to use. Can be one of "ward.D", "ward.D2", "single", "complete", "average", "mcquitty", "median", or "centroid". Defaults to "ward.D".
deepSplit	Logical or integer value from 0 to 4 indicating how fine-grained the clusters should be with larger values giving increased cluster structure. Defaults to FALSE.

Details

Extracting meta-communities allows the user to explore community relatedness and structure at higher levels. Community relatedness is calculated using the Jaccard coefficient and the number of nodes that community i and j share:

S(i,j)=|intersect(i,j)|/|union(i,j)|

A hierarchical clustering dendrogram is generated based on the above distance metric and then an algorithm is used to automatically extract clusters (Langfelder et al. 2008). This function works best on large networks with a large number of link communities.

Value

An object of class linkcomm, which is a list containing the following components:

numbers	An integer vector with the number of edges, nodes, and communities.
hclust	An object of class hclust, which contains information about the hierarchical clustering of links.
pdmax	A numerical value indicating the height of the dendrogram at which the partition density is maximised.
pdens	A numerical matrix with 2 columns; the first is the heights at which clusters appear and the second is the partition density.
nodeclusters	A data frame consisting of 2 columns; the first contains node names, and the second contains single community IDs for each node. All communities and their nodes are represented, but not necessarily all nodes.
clusters	A list of integer vectors containing the link IDs that belong to each community. Community IDs are the numerical position of the communities in the list.
edges	A data frame with 3 columns; the first two contain nodes that interact with each other, and the third is an integer vector of community IDs indicating community membership for each link.
numclusters	A named integer vector. Names are node names and integer values are the number of communities to which each node belongs.
clustsizes	A named integer vector. Names are community IDs and integer values indicate the number of nodes that belong in each community.
igraph	An object of class igraph. The network is represented here as an igraph object.
edgelist	A character matrix with 2 columns containing the nodes that interact with each other.
directed	Logical indicating whether the network is directed.
bipartite	Logical indicating whether the network is bi-partite.

Author(s)

Alex T. Kalinka [e-mail]

References

Kalinka, A.T. and Tomancak, P. (2011). linkcomm: an R package for the generation, visualization, and analysis of link communities in networks of arbitrary size and type. Bioinformatics 27, 2011-2012.

Langfelder, P., Zhang, B., and Horvath, S. (2008) Defining clusters from a hierarchical cluster tree: the Dynamic Tree Cut package for R. Bioinformatics 24, 719-720.

See Also

getClusterRelatedness

Examples

## Generate graph and extract link communities.
g <- swiss[,3:4]
lc <- getLinkCommunities(g)

## Produce meta-communities.
## Not run: mc <- meta.communities(lc)

alextkalinka/linkcomm documentation built on Feb. 11, 2021, 4:53 a.m.