nem.calcSignificance {nem}R Documentation

Statistical significance of network hypotheses

Description

Assess statistical significance of a network hypothesis compared to a random, an empty and a fully connected network.

Usage

        nem.calcSignificance(D, Phi, likelihood, N=1000, seed=1, type="mLL", Pe=NULL, para=NULL, hyperpara=NULL, delta=1, selEGenes=FALSE)

Arguments

D data matrix with experiments in the columns (binary or continious)
Phi network hypothesis (adjacency matrix)
likelihood the network's marginal log-likelihood
N number of random networks to sample
seed random seed
type mLL or FULLmLL or CONTmLL or CONTmLLDens or CONTmLLRatio
Pe prior of effect reporter positions in the phenotypic hierarchy (same dimension as D)
para vector of length two: false positive rate and false negative rate for non-binary data. Used by mLL
hyperpara vector of length four: used by FULLmLL() for binary data
delta regularization parameter for automated E-gene subset selection (CONTmLLRatio only)
selEGenes automated E-gene subset selection (includes tuning of delta for CONTmLLRatio)

Details

Given data, N random network hypotheses from a null distribution are drawn as follows: For each S-gene $S_k$ we randomly choose a number o of outgoing edges between 0 and 3. We then select o S-genes having at most 1 ingoing edge, connected $S_k$ to them and transitively closed the graph. For all random network hypotheses it is counted, how often their likelihood is bigger than that of the given network. This yields an exact p-value.

Comparison of the likelihood of the given network to an empty and a fully connected gives us Bayes factors, which are also returned by the method.

Value

p.value p-value of the network according to the null hypothesis that it is random
BFzero marginal log-likelihood difference (weight of evidence) to the empty network
BFcomplete marginal log-likelihood difference (weight of evidence) to the fully connected network

Author(s)

Holger Froehlich

See Also

nem.bootstrap, nem

Examples

## Not run: 
   data("BoutrosRNAi2002")
   D <- BoutrosRNAiDiscrete[,9:16]
   p <- c(.13,.05)
   res = nem(D, para=p) # get best network
   nem.calcSignificance(D,as(res$graph,"matrix"),res$mLL, para=p) # assess its statistical significance
## End(Not run)

[Package nem version 2.4.0 Index]