### R code from vignette source 'vignettes/stepwiseCM/inst/doc/stepwiseCM.Rnw' ################################################### ### code chunk number 1: chunk1 ################################################### # load the full Central Nervous System cancer data library(stepwiseCM) data(CNS) ################################################### ### code chunk number 2: chunk2 ################################################### # gain the prediction labels of the training set data(CNS) attach(CNS) #Note that columns of the train set should corresponding the sample and rows #corresponding the feature. train.exp <- mrna[, 1:30] train.cli <- t(cli[1:30, ]) train.label <- class[1:30] test.exp <- mrna[, 31:60] test.cli <- t(cli[31:60, ]) pred.exp <- Classifier(train=train.cli, test=test.cli, train.label=train.label, type="RF", CVtype="k-fold", outerkfold=2, innerkfold=2) pred.cli <- Classifier(train=train.exp, test=test.exp, train.label=train.label, type="plsrf_x", CVtype="k-fold", outerkfold=2, innerkfold=2) # Classification accuracy of the training set from clinical data is: sum(pred.cli$P.train == train.label)/length(train.label) # Classification Accuracy of the training set from expression data is: sum(pred.exp$P.train == train.label)/length(train.label) ################################################### ### code chunk number 3: chunk3 ################################################### train.exp <- mrna[, 1:40] train.cli <- t(cli[1:40, ]) train.label <- class[1:40] test.cli <- t(cli[41:60, ]) prox <- Proximity(train.cli=train.cli, train.label=train.label, test.cli=test.cli, train.gen=train.exp, N = 2) prox1 <- Proximity(train.cli=train.cli, train.label=train.label, test.cli=test.cli, train.gen=train.cli, N = 2) #check the range of proximity values from two different data types par(mfrow=c(1,2)) plot(sort(prox1$Prox.gen[1, ][-1], decreasing=TRUE), main="clinical", xlab="Rank", ylab="Proximity", ylim=c(0,1)) plot(sort(prox$Prox.gen[1, ][-1], decreasing=TRUE), main="expression", xlab="Rank", ylab="Proximity", ylim=c(0,1)) ################################################### ### code chunk number 4: chunk4 ################################################### train.exp <- mrna[, 1:30] train.cli <- t(cli[1:30, ]) train.label <- class[1:30] test.cli <- t(cli[31:60, ]) pred.cli <- Classifier(train=train.cli, test=c(), train.label=train.label, type="RF", CVtype="k-fold", outerkfold=2, innerkfold=2) pred.exp <- Classifier(train=train.exp, test=c(), train.label=train.label,type="GLM_L1", CVtype="k-fold", outerkfold=2, innerkfold=2) prox <- Proximity(train.cli=train.cli, train.label=train.label, test.cli=test.cli, train.gen=train.exp, N = 2, Parallel = FALSE) RS <- RS.generator(pred.cli=pred.cli$P.train, pred.gen=pred.exp$P.train, train.label=train.label, prox.gen=prox$Prox.gen, prox.cli=prox$Prox.cli, type = "both") #observe the differences by ranking the RS values order(RS[, 1]) # from the ranking approach order(RS[, 2]) # from the proximity approach ################################################### ### code chunk number 5: chunk5 ################################################### tr.exp <- mrna[, 1:40] tr.cli <- t(cli[1:40, ]) tr.label <- class[1:40] te.exp <- mrna[, 41:60] te.cli <- t(cli[41:60, ]) te.label <- class[41:60] result <- Curve.generator(train.cli=tr.cli, train.gen=tr.exp, train.label=tr.label, test.cli= te.cli, test.gen=te.exp, test.label=te.label, type=c("RF", "TSP"), RStype = "rank", Parallel = FALSE, CVtype = "k-fold", outerkfold = 2, innerkfold = 2, N = 2, plot.it=TRUE) ################################################### ### code chunk number 6: chunk6 ################################################### tr.cli <- t(cli[1:40, ]) te.cli <- t(cli[41:60, ]) tr.gen <- mrna[, 1:40] te.gen <- mrna[, 41:60] tr.label <- class[1:40] te.label <- class[41:60] result <- Curve.generator(train.cli=tr.cli, train.gen=tr.exp, train.label=tr.label, test.cli= te.cli, test.gen=te.exp, test.label=te.label, type=c("RF", "plsrf_x"), RStype = "rank", Parallel = FALSE, CVtype = "k-fold", outerkfold = 2, innerkfold = 2, N = 2, plot.it = FALSE) A <- Step.pred(result, test.cli, 30)