dna (version 2.1-2)

test.modular.structure: Test for differential modular structures

Description

Tests for differential modular structures between two networks using PLS scores.

Usage

test.modular.structure(X1,X2,scores="PLS",min.module.size=5,epsilon=.5,
num.permutations=1000,check.networks=TRUE,...)

Arguments

X1

network 1 with genes in columns and samples in rows.

X2

network 2 with genes as columns and samples in rows.

scores

type of connectivity score to be used. Either one of the built-in methods ("PLS", "PC", "RR", or "cor") can be used or a user-defined method can be supplied.

min.module.size

minimum module size parameter.

epsilon

threshold parameter.

num.permutations

the number of random permutations.

check.networks

indicates whether get.common.networks is used to preprocess the networks before the test is performed.

...

additional arguments for scores.

Value

results

result of test (of class resultsModTest).

References

Gill, R., Datta, S., and Datta, S. (2010) A statistical framework for differential network analysis from microarray data. BMC Bioinformatics, 11, 95.

Examples

# NOT RUN {
# small example illustrating test procedures
set.seed(12345)
X1=rbind(
c(2.5,6.7,4.5,2.3,8.4,3.1),
c(1.2,0.7,4.0,9.1,6.6,7.1),
c(4.3,-1.2,7.5,3.8,1.0,9.3),
c(9.5,7.6,5.4,2.3,1.1,0.2))
colnames(X1)=paste("G",1:6,sep="")

X2=rbind(
c(4.5,2.4,6.8,5.6,4.5,1.2,4.5),
c(7.6,9.0,0.1,3.4,5.6,5.5,1.2),
c(8.3,4.5,7.0,1.2,4.3,3.7,6.8),
c(3.4,1.1,6.9,7.2,3.1,0.9,6.6),
c(3.4,2.2,1.3,5.5,9.8,6.7,0.6))
colnames(X2)=paste("G",8:2,sep="")

# perform a test for modular structure using a minimum module size of 2
# and threshold of .5 with PLS connectivity scores
## Not run: tms=test.modular.structure(X1,X2,min.module.size=2)
## Not run: summary(tms)

# extract results for a test of modular structure
## Not run: results.tms=get.results(tms)
## Not run: results.tms

# perform a test for modular structure using a minimum module size of 2
# and threshold of .5 with PLS connectivity scores without rescaling the data,
# symmetrizing the scores, or rescaling the scores based on 10000 permutations
## Not run: tms2=test.modular.structure(X1,X2,scores="PLS",min.module.size=2,
## num.permutations=10000,rescale.data=FALSE,symmetrize.scores=FALSE,
## rescale.scores=FALSE)
## Not run: summary(tms2)

# perform a test for modular structure using a minimum module size of 2
# and threshold of .5 with correlation connectivity scores
## Not run: tms3=test.modular.structure(X1,X2,scores="cor",min.module.size=2)
## Not run: summary(tms3)

# perform a test for modular structure using a minimum module size of 3
# and threshold of .7 with principal components regression connectivity scores
## Not run: tms4=test.modular.structure(X1,X2,scores="PC",min.module.size=3,
## epsilon=.7)
## Not run: summary(tms4)

# perform a test for modular structure using a minimum module size of 2
# and threshold of .9 with ridge regression connectivity scores with 
# rescaled data, symmetrized and rescaled scores and a penalty parameter 
# equal to 3
## Not run: tms5=test.modular.structure(X1,X2,scores="RR",min.module.size=2,
## epsilon=.5,rescale.scores=TRUE,lambda=3)
## Not run: summary(tms5)

# perform a test for modular structure using a minimum module size of 2 and 
# threshold of .9 with custom ridge regression connectivity scores with 
# centered and rescaled data and symmetrized and rescaled scores
## Not run: ourRR=function(X,y,lambda=3){
## solve(t(X)%*%X+lambda*diag(ncol(X)))%*%t(X)%*%y}
## Not run: ourRRnet=function(X){gennet(X,f=ourRR,recenter.data=TRUE,
## rescale.data=TRUE,symmetrize.scores=TRUE,rescale.scores=TRUE)}
## Not run: tms6=test.modular.structure(X1,X2,scores=ourRRnet,
## min.module.size=2,epsilon=.9)
## Not run: summary(tms6)
# }