c3mtc: 'c3mtc' gene regulatory network inference using c3net with...

Description Usage Arguments Value Author(s) References See Also Examples

View source: R/c3mtc.R


We present in the following the inference method C3NET introduced in Altay (2010a) in a modified form to obtain a more efficient implementation. Briefly, C3NET consists of three main steps. First, mutual information values among all gene pairs are estimated. Second, an extremal selection strategy is applied allowing each of the p genes in a given dataset to contribute at most one edge to the inferred network. That means we need to test only p different hypotheses and not p(p-1)/2. This potential edge corresponds to the hypothesis test that needs to be conducted for each of the p genes. Third, a multiple testing procedure is applied to control the type one error.

In order to determine the statistical significance of the mutual information values between genes we test for each pair of genes the following null hypothesis.

H_0^I: The mutual information between gene i and j is zero.

Because we are using a nonparametric test we need to obtain the corresponding null distribution for H_0^I from a randomization of the data.

The formulated null hypothesis is performed by permuting the sample and gene labels for all genes of the entire expression matrix at once. The vector of the mutual information null distribution is obtained from repeated randomizations for a given number of iterations.


c3mtc(dataset, null=NULL, mtc=TRUE, adj="bonferroni", alpha=0.05, nullit=NA,
estimator="pearson", disc="none", adjacency=FALSE, igraph=TRUE)



gene expression dataset where rows define genes and columns samples


nullit defines the size of the generated null distribution vector used for hypothesis testing of significant edges inferred by c3net. The null distribution of mutual information is generated from sample and gene label randomization.

default number of iterations: nullit=ceiling(10^5/(((genes*genes)/2)-genes)) genes: number of genes


minet package (continuous estimators) "pearson", "spearman", "kendall", "spearman"

minet package (discrete estimators) "mi.empirical", "mi.mm","mi.sg","mi.shrink"

c3net gaussian estimator (pearson) "gaussian"

bspline requires installation of "mis_calc" "bspline"


only required for discrete estimators (minet package) "equalfreq", "equalwidth"


consider multiple hypothesis testing for edges inferred by c3net


if mtc==TRUE default multiple hypothesis testing procedure for c3net inferred edges using "bonferroni" (default)

alternatively use "holm", "hochberg", "hommel", "bonferroni", "BH", "BY","fdr", "none" (see ?p.adjust())


significance level for mtc after multiple hypothesis testing correction


return an adjacency matrix


return igraph object


If NULL a null distribution vector is generated from a sample label and gene label permutation of the gene expression matrix. For the ensemble inference of one dataset an external null distribution vector is suggested for decreasing running time.


'c3mtc' returns a gene regulatory network formated as adjacency matrix, as weighted matrix where the edge weights are defined by the corresponding mutual information values or as undirected weighted or unweighted igraph object.


de Matos Simoes R, Emmert-Streib F.


Altay G, Emmert-Streib F. Inferring the conservative causal core of gene regulatory networks. BMC Syst Biol. 2010 Sep 28;4:132. PubMed PMID: 20920161; PubMed Central PMCID: PMC2955605.

de Matos Simoes R, Emmert-Streib F. Bagging statistical network inference from large-scale gene expression data. PLoS One. 2012;7(3):e33624. Epub 2012 Mar 30. PubMed PMID: 22479422; PubMed Central PMCID: PMC3316596.

de Matos Simoes R, Emmert-Streib F. Influence of statistical estimators of mutual information and data heterogeneity on the inference of gene regulatory networks. PLoS One. 2011;6(12):e29279. Epub 2011 Dec 29. PubMed PMID: 22242113; PubMed Central PMCID: PMC3248437.

See Also




bc3net documentation built on May 30, 2017, 5:22 a.m.