Main - Probit selection prior

Share:

Description

Perform MCMC iterations of the model described in the reference.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
iBATProbit(Y, X, distance, disfix, intercept=1, xi,
R=-1, tran, mu, sigma=((rgamma(4,1,1))^(-0.5)),
cmu=1/1000000, c=10, delta=3, d, alpha0=2.32,
alpha1=1, deltak=c(-1,0,0.58,1), tauk=c(1,1,1,2),
upp_bounds=c(-0.1, 0.1, 0.73, Inf),
low_bounds=c(-Inf, -0.1, 0.1, 0.73),
alpha_IG=c(1,1,1,1), beta_IG=c(1,1,1,1),
low_IG=c(0.41,0.41,0.41,1), a=c(1,1,1,1),
niter=500000, burnin=200000, Cout=1000,
phi=0.5, pR=0.4, selectioncgh=-1, pXI=0.6)

Arguments

Y

Matrix of gene expression data

X

Matrix of CGH data

distance

Vector of distance between CGH probes

disfix

Length of the chromosome under investigation

intercept

If set to one an intercept is included in the regression model

xi

Initialized matrix of latent states

R

Initialized association matrix in a vector form. Default set to -1, that automatically creates a vector with all the positions set to zero

tran

Initialized transition matrix

mu

Initialized state specific mean vector

sigma

Initialized state specific standard deviation vector

cmu

Parameter that controls the variance of the prior on the intercept

c

Parameter that determines the shrinkage in the model

delta

Parameter of the Inverse-Gamma prior on the error variance

d

Parameter of the Inverse-Gamma prior on the error variance

alpha0

Baseline intercept of the selection prior

alpha1

Parameter that regulates the strength of the spatially informed dependence

deltak

Vector of mean of the prior on the state specific mean

tauk

Vector of sd of the prior on the state specific mean

upp_bounds

Vector of upper bounds of the prior on the state specific mean

low_bounds

Vector of lower bounds of the prior on the state specific mean

alpha_IG

Parameter of the prior on the state specific standard deviation

beta_IG

Parameter of the on the state specific standard deviation

low_IG

Truncation of the prior on the state specific standard deviation

a

Vector of parameters of the prior on the transition matrix

niter

Number of Monte Carlo Markov Chain iteration

burnin

Burn-in

Cout

Print the number of iterations ran every Cout iterations

phi

Probability of an A/D step

pR

Parameter of the distribution used to select the rows to be updated at every MCMC iteration

selectioncgh

Number of samples not in neutral state in order to consider a CGH as a potential candidate for association with gene expression. Default set to -1 that automatically set it to 10% of the samples

pXI

Parameter of the distribution used to select the rows to be updated at every MCMC iteration

Value

The output consists of an R list composed by 4*niter+3 objects, where niter is the number of MCMC iterations. The first niter objects of the list are vectors, each containing the positions of the association matrix set to one, at the corresponding MCMC iteration. Each of the following niter objects of the list are the transition matrices at the corresponding MCMC iteration, while the third and the fourth set of niter objects are the vectors of state specific mean and state specific variance, respectively. The last three objects of the list consist of three matrices counting the number of times the corresponding latent state has been set to 1,3 and 4, respectively.

Author(s)

Alberto Cassese

References

Cassese A, Guindani M, Vannucci M. A Bayesian integrative model for genetical genomics with spatially informed variable selection. Cancer Informatics.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
## Not run: 
data(TCGA_lung)

Y <- TCGA_lung$Affy
X <- TCGA_lung$aCGH
distance <- TCGA_lung$distance
disfix <- 199446827
xi <- InitXi(X)
tran <- Tran(xi)
mu <- InitMu()
d=0.2587288

Y <- Center(Y)

res <- iBATProbit(Y=Y,X=X,distance=distance,disfix=disfix,xi=xi,tran=tran,mu=mu,d=d)

summRes <- Inference(res,G=dim(Y)[[2]],M=dim(X)[[2]],niter=niter,burnin=bi,threshold=0.5)

## End(Not run)