load('dataLeukemia.RData')

names(data[.1:10])
data$Leukemia.class = as.numeric(data$Leukemia.class)
data = data[.-c(1,3)]
dataSel = data[data$Leukemia.class ==11 |data$Leukemia.class ==15,]
table(data_sel$Leukemia.class)

indexTrain = indexTest = list()
indexAll = 1:391
indexTest[[1]]= c(1:66,199:262)
indexTrain[[1]]= indexAll[-c(1:66,199:262)]
indexTest[[2]]= c(67:132,263:327)
indexTrain[[2]]= indexAll[-c(67:132,263:327)]
indexTest[[3]]= c(133:198,328:391)
indexTrain[[3]]= indexAll[-c(133:198,328:391)]

dim(dataSel)

class = dataSel[,1]
dataGenes = dataSel[,-1]

#resultTest = tTest(x =c(dataGenes[,1]), y=class.alternative = c("two.sided"), var.equal = TRUE)
#usunac nadmiarowe dane, te niepotrzebne

tTest = function(m,dataGenes, class){
  resultTest =t.test(x =c(dataGenes[,1]), y=class, alternative = c("two.sided"), var.equal = TRUE)
  resultTestpValue = resultTest$p.value
  return(resultTest$p.value)
}

dataGenes0 = dataGenes[indexTrain[[3]],]
class0 = class[indexTrain[[3]]]

listpValue = list()
for(i in 1:ncol(dataGenes0)){
  listpValue[[i]] = tTest(i,dataGenes0, class0)
}

vecpValue = unlist(listpValue)
nameGenepValue = cbind(names(dataGenes), vecpValue)
nameGenepValueSort = nameGenepValue[order(nameGenepValue[,2]),]
nameGenepValueSortAdj = cbind(nameGenepValueSort[.1], p.adjust(nameGenepValueSort[.2], method = 'BH'))

varImp<- nameGenepValueSortAdj[which(nameGenepValueSort[.2]<0.05),]

resultModel = randomForest:: randomForest(x= dataGenes[indexTrain[[3]]], varImp[1:100,1], y= as.factor(class[indexTrain[[3]]]),
                                          xtest = dataGenes[indexTest[[3]]], varImp[1:100,1], yTest = as.factor(class[indexTest[[3]]]),
                                          nTree = 500, importance = True)

resultModel$importance