mLL {nem}R Documentation

Marginal likelihood of a phenotypic hierarchy

Description

computes the marginal likelihood of observed phenotypic data given a phenotypic hierarchy.

Usage

mLL(Phi,D1,D0=NULL,a=0.15,b=0.05,Pe=NULL,Pm=NULL,lambda=0,type="mLL")

Arguments

Phi an adjacency matrix with unit main diagonal
D1 (i) count matrix for discrete data: phenotypes x genes. How often did we see an effect after interventions? (ii) matrix describing the PROBABILITIES of an effect (iii) matrix discribing the log-LIKELIHOOD of an effect (e.g. log-density matrix, log-odds matrix)
D0 count matrix: phenotypes x genes. How often did we NOT see an effect after intervention? Not used for continious data
a false positive rate: how probable is it to miss an effect? (for count matrix)
b false negative rate: how probable is it to see a spurious effect? (for count matrix)
Pe prior of effect reporter positions in the phenotypic hierarchy
Pm prior on model graph (n x n matrix) with entries 0 <= priorPhi[i,j] <= 1 describing the probability of an edge between gene i and gene j.
lambda regularization parameter to incorporate prior assumptions.
type see nem

Details

It computes the marginal likelihood of a single phenotypic hierarchy. Usually called from within the function score.

Value

mLL marginal likelihood of a phenotypic hierarchy
pos posterior distribution of effect positions in the hierarchy
mappos Maximum aposteriori estimate of effect positions
LLperGene likelihood per E-gene

Author(s)

Holger Froehlich <URL: http://www.dkfz.de/mga2/people/froehlich>, Florian Markowetz <URL: http://genomics.princeton.edu/~florian>

References

Markowetz F, Bloch J, Spang R, Non-transcriptional pathway features reconstructed from secondary effects of RNA interference, Bioinformatics, 2005

See Also

nem, score, FULLmLL

Examples

   data("BoutrosRNAi2002")
   result <- nem(BoutrosRNAiDiscrete[,9:16],type="mLL",para=c(.15,.05))

[Package nem version 2.6.0 Index]