Networks.STD: Standard Algorithm for Bayesian Network Discovery

Description Usage Arguments Details Value References Examples

View source: R/Networks.STD.R

Description

Standard Algorithm for Bayesian Network Discovery

Usage

1
2
3
Networks.STD(pvalue, net, iter = 5000, nburns = 2000, piall = c(0.75, 0.8,
  0.85, 0.9), rhoall = c(0.5, 1, 5, 10, 15), status = FALSE, fit,
  show.steps = 1, showlikelihood = FALSE, likelihood.frequency = 100)

Arguments

pvalue

a vector of p-values obtained from large scale statistical hypothesis testing

net

a n by n network configuration, n is the length of pvalue

iter

number of iterations. The default is 5000

nburns

number of burn-in. The default is 2000

piall

a vector of selections of pi0. The default vector is 0.75, 0.8, 0.85, 0.9. The selections of pi0 should be placed in sequence, from smaller to larger.

rhoall

a vector of selections of rho0 and rho1. The default vector is 0.5, 1, 5, 10, 15. The selections of rho0 and rho1 should be placed in sequence, from smaller to larger.

status

default=FALSE

fit

NULL

show.steps

number default=1

showlikelihood

logical default=FALSE

likelihood.frequency

number default=100

Details

This generic function fits a Bayesian Nonparametric Mixture Model for gene selection incorporating network information (Zhao et al., 2014):

where we define

Index

a_0=(-L_0+1,-L_0+2,...,0) , a_1=(1,2,...,L_1) and the correspondent probability q_0=(q_-L_0+1, q_-L_0+2, ...,q_0), q_1=(q_1, q_2, ..., q_L_1), according to the defination of Discrete(a_k, b_k), for example, Pr(g_i=L_0+2)=q_-L_0+2.

Assumption

We have an assumption that "selected" gene or image pixel should have larger statiscs comparing to "unselected" ones without the loss of generality. In this regard, we set the restriction mu_g<mu_g+1 for g=-L_0+1, -L_0+2,...,L_1.

For this function, The NET-DPM-1, considered as standard function is applied , and more details about the algorithm can be referred from Appnendix B.1 of Zhao et al., 2014

Value

The trace of gi showing the evolution of the Monte Carlo Markov Chain

References

Zhao, Y.*, Kang, J., Yu, T. A Bayesian nonparametric mixture model for gene and gene-sub network selection Annals of Applied Statistics, In press: 2014.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
#' ##Creating the network of 10X10 image
## Not run: 
library(igraph)
library(BayesNetDiscovery)
g <- graph.lattice(length=10,dim=2)
## the input of argument \code{net}
net=as(get.adjacency(g,attr=NULL),"matrix")
##Assign the signal elements with signal intenstion as normal distribution N(1,0.2). 
While noise is set as N(0,0.2) 
newz=rep(0,100)
for (i in 3:7)
{
 newz[(i*10+3):(i*10+7)]=1
}
testcov<-0
for(i in 1:100){
 if(newz[i]==0){
   testcov[i]<-rnorm(1,mean=0,sd=0.2)
 
 }else{
  testcov[i]<-rnorm(1,mean=1,sd=0.2)
   
 }
}
##The profile of the impage
image(matrix(testcov,10,10),col=gray(seq(0,1,length=255)))
##Transform the signals into pvalue form and begin identification
pvalue=pnorm(-testcov)
total=Networks.STD(pvalue,net,iter=5000,piall=c(0.8, 0.85, 0.9, 0.95),
rhoall=c(0.5,1,5,10,15))

## End(Not run)

BANFF documentation built on May 29, 2017, 11:59 a.m.

Related to Networks.STD in BANFF...