computeModules: computeModules

Description

This function takes a bipartite weighted graph and computes modules by applying Newman's modularity measure in a bipartite weighted version to it. metaComputeModules re-runs the algorithm several times, returning the most modular result, to stabilise modularity computation.

Usage

computeModules(web, method="Beckett", deep = FALSE, deleteOriginalFiles = TRUE, 
	steps = 1000000, tolerance = 1e-10, experimental = FALSE, forceLPA=FALSE)
	
metaComputeModules(moduleObject, N=5, method="Beckett", ...)

Value

An object of class "moduleWeb" containing information about the computed modules. For details, please refer to the corresponding documentation file.

Arguments

web: web is the matrix representing the weighted bipartite graph (as an example, see e.g. web small1976 in this package). This matrix can be binary (i.e. consist only of 0s and 1s), in which case the output will be Newman's (2006) modularity.
method: Choice between the algorithm(s) provided by Stephen Beckett (2016) or Dormann & Strauss (2016) (method="DormannStrauss"). Defaults to the much faster and in the majority of cases better algorithm of Beckett. (Note the optional argument forceLPA to use his slightly inferior but even faster pure LPA algorithm.)
deep: If deep is set to FALSE (default), a flat clustering is computed, otherwise submodules are identified recursively within modules. Works only with method="DormannnStrauss".
deleteOriginalFiles: If deleteOriginalFiles is set to TRUE (default), the files mentioned above in the description are deleted from the hard drive disk, otherwise not. Applies only to method="DormannnStrauss".
steps: steps is the number of steps after which the computation of modules stops if no better division into modules than the current one can be found. Applies only to method="DormannnStrauss".
tolerance: How small should the difference between MCMC-swap results be? At some point computer precision fluctuations make the algorithm fail to converge, which is why we choose a (very low) defaults of 1E-10. Applies only to method="DormannnStrauss".
experimental: Logical; using an undescribed and untested version for which no detail is available? (We suggest: not yet.) Applies only to method="DormannnStrauss".
moduleObject: Output from running computeModules.
forceLPA: Logical; should the even faster pure LPA-algorithm of Beckett be used? DIRT-LPA, the default, is less likely to get trapped in a local minimum, but is slightly slower. Defaults to FALSE. Applies only to method="Beckett".
N: Number of replicate runs; defaults to 5. Not really required for method="Beckett", which starts in different places anyway.
...: Arguments passed on to computeModules, which is called internally.

Author

Rouven Strauss, with fixes by Carsten Dormann and Tobias Hegemann; modified to accommodate Beckett's algorithm by Carsten Dormann

References

Beckett, S.J. 2016 Improved community detection in weighted bipartite networks. Royal Society open science 3, 140536.

Dormann, C. F., and R. Strauß. 2014. Detecting modules in quantitative bipartite networks: the QuanBiMo algorithm. Methods in Ecology & Evolution 5 90--98 (and arXiv [q-bio.QM] 1304.3218.)

Liu X. & Murata T. 2010. An Efficient Algorithm for Optimizing Bipartite Modularity in Bipartite Networks. Journal of Advanced Computational Intelligence and Intelligent Informatics (JACIII) 14 408--415.

Newman M.E.J. 2004. Physical Review E 70 056131

Newman, M.E.J. 2006. Modularity and community structure in networks. Proceedings of the National Academy of Sciences of the United States of America, 103, 8577–-8582.

Examples

Run this code

	if (FALSE) {
		data(small1976)
		(res <- computeModules(small1976)) 
		plotModuleWeb(res)
		
		# slow:
		res2 <- metaComputeModules(small1976, method="DormannStrauss")
		res2
	}

Run the code above in your browser using DataLab