Implements the Leiden algorithm via an R interface
Since October 2020, the R package igraph contains the function cluster_leiden()
implemented by Vincent Traag (@vtraag). The usage of this function is detailed in the igraph documentation here. We recommend users use this function.
There is still no R package which entirely encompasses all of the functionality of the original Python/C++ implementation here from https://github.com/vtraag/leidenalg. We hope interested developers could use this package as a starting point for creating an R package which mirrors the full available functionality of the Python package. See here for details.
The Leiden algorithm is an iterative community detection algorithm on networks---the algorithm is designed to converge to a partition in which all subsets of all communities are locally optimally assigned, yielding communities guaranteed to be connected.
The algorithm was written to improve upon defects of the Louvain algorithm. Consequently, the Leiden algorithm is faster, scales well, and can be run on graphs of millions of nodes (as long as they can fit in memory).
The basic steps are:
For details on the algorithm, see "From Louvain to Leiden: guaranteeing well-connected communities" Traag, Waltman, van Eck. Sci Rep 9, 5233 (2019). https://doi.org/10.1038/s41598-019-41695-z
For the original implementation in C++ with python bindings, see: https://github.com/vtraag/leidenalg
To install the stable version from CRAN, use:
install.packages('leidenAlg')
To install the latest version, use:
install.packages('devtools')
devtools::install_github('kharchenkolab/leidenAlg', build_vignettes = TRUE)
Note that this package depends on igraph, which requires various libraries to install correctly e.g. libxml2
. Please see the installation instructions at that page for more details, along with the README here.
Debian-based users of Linux can install the required packages via:
sudo apt-get update
sudo apt-get install libxml2-dev libgmp-dev libglpk-dev
For users of Red Hat distributions, use the following command to install the required packages:
sudo yum update
sudo yum install libxml2-devel gmp-devel glpk-devel
For Mac OS, the commands with the Homebrew package manager are as follows:
brew update
brew install libxml2 glpk gmp
Note: For Mac OS users, there is a guide for troubleshooting here if issues arise.
find_partition()
: Finds the optimal partition using the Leiden algorithm.
find_partition_with_rep()
: Finds the optimal partition using the Leiden algorithm with replicate starts
leiden.community()
: Detects communities using Leiden algorithm, output as fakeCommunities
class for downstream use.
rleiden.community()
: Recursive leiden communities, constructs an n-step recursive clustering, using leiden.community.detection. Returns a fakeCommunities
object that has methods membership(), without dendrogram.
as.dendrogram()
: Returns pre-calculated dendrogram from "fakeCommunities"
object
membership()
: Returns pre-calculated membership factor from "fakeCommunities"
object
If you find leidenAlg
useful for your publication, please cite:
Peter Kharchenko, Viktor Petukhov, Yichen Wang, and Evan Biederstedt (2023).
leidenAlg: Implements the Leiden Algorithm via an R Interface. R
package version 1.1.4. https://github.com/kharchenkolab/leidenAlg