MMG-package {MMG}R Documentation

Mixture Model on Graphs

Description

MMG is a Mixture Model which can integrate the structure of a network in the statistical analysis of data.

This implementation MMG assumes the existence of three classes of genes/proteins/etc. within a network: down-regulated, up-regulated, or unchanged. The underlying data are log-ratios, hence relative, measurements.

The package aims at identifying clusters of genes/proteins/etc. that behave consistently along the network's pathways. This is done by implementing the Bayesian model described in the Reference Sanguinetti et al. (2008), by running a Gibbs sampler, and by cutting the graph.

The Gibbs sampler is run by MMG.compute. To cut the graph is done by MMG.cut.graph. Finally, MMG.make.dot conveniently produces a DOT file, a file whose format can be used for visualation using various softwares (including GraphViz http://www.graphviz.org/ - see also the package Rgraphviz).

Details

Package: MMG
Type: Package
Version: 1.2.2
Date: 2008-05-20
License: GPL-3+

Author(s)

Josselin Noirel, based on an original implementation by Guido Sanguinetti (http://www.dcs.shef.ac.uk/~guido/).

Maintainer: Josselin Noirel <j.noirel@sheffield.ac.uk>

References

Sanguinetti, Noirel, and Wright., MMG: a probabilistic tool to identify submodules of metabolic pathways, Bioinformatics (2008)

Examples

## Not run: 
r <- MMG.compute(file.name = "NostocData/R_net.dat",
                 steps = 100000, burn.in = 1000,
                 sigma = 0.3, alpha = 1)
## End(Not run)
## Not run: n <- r$dat$n.nodes
## Not run: 
s <- MMG.cut.graph(r, descriptions = "NostocData/R_descr.dat",
                   method = "THRESHOLD", threshold = 0.15, select = "UP")
## End(Not run)
## Not run: l <- (1:n)[s$components != 0]
## Not run: 
MMG.make.dot(r, file.name = "nostoc.dot", selection = l, type = "UNDIRECTED",
             rem.loops = TRUE)
## End(Not run)

[Package MMG version 1.4.0 Index]