get.phi.final: Construct final network from GA or mcmc result.

In ddepn: Dynamic Deterministic Effects Propagation Networks: Infer signalling networks for timecourse RPPA data.

Description

get.phi.final takes the output of netga and constructs a final network from the population, get.phi.final.mcmc takes a list containing the returned lists of mcmc_ddepn and calculates a final network for each result list.

Usage

get.phi.final(lst, th = 0.8)
get.phi.final.mcmc(retlist,maxiterations,prob=.333,qu=.99999)

Arguments

lst	Output list from netga
th	Double in [0;1]. Threshold for inclusion of an edge into the final network.
retlist	A list returned by MCMC inference.
maxiterations	Integer. Number of MCMC sampling iterations.
prob	Double \in [0;1].Success probability for binomial density.
qu	Double \in [0;1]. Quantile of the binomial distribution, used as significance cutoff for edge inclusion.

Details

get.phi.final Takes the population P from the GA resultlist and returns the list with the element lst$phi replaced by the new final network. lst$weights only contains the weights with lst$weights > th.
get.phi.final.mcmc Takes a list of MCMC samplings returned by mcmc_ddepn and extracts a final network for each result. It is assumed that any type of edge is expected to be found with probability 1/3=0.33, thus for maxiterations samplings we expect the probability of seeing exactly k edges of a certain type follows a binomial distribution B(maxiterations, prob, k). An edge is included, if the probability of seeing more than the observed number of occurrences is less or equal than 1-qu, i.e. P(k>Kobserved) <= 1-qu.

Value

Result list as in netga, with replaced phi and weights elements.

Author(s)

Christian Bender

See Also

netga

Examples

## Not run: 
## load package
library(ddepn)
## sample a network
n <- 6
signet <- signalnetwork(n=n, nstim=2, cstim=0, prop.inh=0.2)
phit <- signet$phi
stimuli <- signet$stimuli
## sample data
dataset <- makedata(phit, stimuli, mu.bg=1200, sd.bg=400, mu.signal.a=2000, sd.signal.a=1000)
## Genetic algorithm
ret1 <- ddepn(dataset$datx, phiorig=phit, inference="netga",
              maxiterations=30, p=15, q=0.3, m=0.8, P=NULL,
              usebics=TRUE) 

plotrepresult(ret1,outfile=NULL)
ret2 <- get.phi.final(ret1, th=0.9)
plotrepresult(ret2,outfile=NULL)

## mcmc
maxiterations <- 300
## use original network as prior matrix
## reset all entries for inhibiting edges 
## to -1
B <- phit
B[B==2] <- -1

ret3 <- ddepn(dataset$datx,phiorig=phit,
      inference="mcmc", usebics=FALSE,
      maxiterations=maxiterations, burnin=100, lambda=0.01, B=B)

plotrepresult(ret3$samplings[[1]],outfile=NULL)
ret4 <- get.phi.final(ret3$samplings[[1]],th=0.9)
plotrepresult(ret4,outfile=NULL)


## End(Not run)

ddepn documentation built on May 2, 2019, 4:42 p.m.