Code

library(ripe)
library(limma)
data(DREAM4)

design <- model.matrix(~ 0+factor(rep(1:11,each=5)))
colnames(design) <- unique(colnames(pertData))
fit <- lmFit(pertData,design)

contrast.matrix <- makeContrasts(K01-Wt, K02-Wt, K03-Wt,K04-Wt,K05-Wt,K06-Wt,K07-Wt,K08-Wt,K09-Wt,K10-Wt, levels=design)
fit2 <- contrasts.fit(fit, contrast.matrix)
fit2 <- eBayes(fit2) # moderated t-statistics

pvalMatrix<-t(fit2$p.value)
diag(pvalMatrix)<-NA

infMatrix <- influencePlot(pvalMatrix,c(1e-6,seq(0.001,0.3,0.005)))

ords <- getOrderings(infMatrix,1000,25,name="insilico_size10_1")

DAGScores <- DAGsFromOrders(ssData,ords,infWgt=infMatrix)

nbest <- ceiling(length(DAGScores$lklhds)*c(0.1,0.2,0.3))
out <- consensusNet(DAGScores,nbest=nbest)
adjMatrix <- out[[3]]$adjMat
adjMatrix

measures<-accMeasures(adjMatrix,goldAdjMatrix)
measures

g <- graph.adjacency(adjMatrix)
V(g)$name <- colnames(adjMatrix);V(g)$label <- V(g)$name
plot(g)

data(Layered)
m<-nrow(infMatr)
p<-ncol(infMatr)
o <- getOrderings(infMatr[,1:m],1000,50,name="LayeredNet")
DAGlkhds<-DAGsFromOrders(exprDat,o,infWgt=infMatr,infConf=0.8)
res<-consensusNet(DAGlkhds,nbest=5,thresh=0.25)
estAdj <- res[[1]]$adjMat
a1 <- accMeasures(estAdj,goldAdj)
a2 <- accMeasures(rbind(estAdj,matrix(0,59,100)),rbind(goldAdj,matrix(0,59,100)))