library(dplyr)
library(caret)
library(randomForest)

#x<-sample.int(sizeData1, size= sizeData1, replace=FALSE)
#dataOneClass1<-dataOneClass1[index.permut1,]
#folds1 <-cut(seq(1,length(dataOneClass1[,1])),breaks=k_cross,labels=FALSE)
load('dataLeukemia.RData')
dim(data)

data=data[,-c(1,3)]
View(data[1:50,1:10])

table(data$Leukemia.class)

data0=data[data$Leukemia.class == 'c-ALL/Pre-B-ALL without t(9;22)' | data$Leukemia.class =='MDS', ]  # call-198 , MDS - 193

dim(data0)
data0$Leukemia.class= as.numeric(data0$Leukemia.class)
print(data0$Leukemia.class)
View(data0)

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

result_test=t.test()


class = data0[,1]
data_genes= data0[,-1]

fun_ttest = function(m,data_genes,class){
  resutlt_test =t.test(x= c(data_genes[,m+1]),y=class,alternative = c("two.sided"),var.equal = TRUE)
  result_test_pvalue =resutlt_test$p.value
  return(result_test_pvalue)
}

list_pValue=list()
for(i in 1:ncol(data_genes)){
  list_pValue[[i]]= fun_ttest(i,data_genes,class)
}

vec_pValue = unlist(list_pValue)
name_gene_pValue = cbind(names(data_genes),vec_pValue)
name_gene_pValue_sort = name_gene_pValue[order(name_gene_pValue[,2]),]
name_gene_pValue_sort_adj = cbind(name_gene_pValue_sort[,1],p.adjust(name_gene_pValue_sort[,2],method ='BH'))
var.imp<-name_gene_pValue_sort_adj[which(name_gene_pValue_sort_adj[,2]<0.05),]
dim(var.imp)

result_model= randomForest::randomForest(x=data_genes[index_train[[1]],1:100], y=as.factor(class[index_train[[1]]]),  xtest=data_genes[index_train[[1]],1:100],
                                         ytest=as.factor(class[index_train[[1]]]),
                                         ntree=500,
                                         importance=TRUE,  do.trace=TRUE)


list_test=list();
list_test[[1]]=c(data0[1:66],data0[199:262])
list_test[[2]]=c(data0[67:132],data0[263:327])
list_test[[3]]=c(data0[133:198],data0[328:391])
list_train=list();
list_train[[1]]=data0[-c(data0[1:66,],data0[199:262,])]
list_train[[2]]=data0[-c(data0[67:132,],data0[263:327,])]
list_train[[3]]=data0[-c(data0[133:198,],data0[328:391,])]



fx <-function(k,data){
  dataclass1<-as.data.frame(data[data[,1]==0,1,3])
  dataclass2<-as.data.frame(data[data[,1]==0,2,3])
  size1 = nrow(dataclass1)
  size2 = nrow(dataclass2)
  index.permut1 = sample.int(size1, size=dataclass1,replace = FALSE)
  index.permut2 = sample.int(size2, size=dataclass2,replace = FALSE)
  dataclass1<-dataclass1[index.permut1,]
  dataclass2<-dataclass2[index.permut2,]
  f1<-cut(seq(1,length(dataclass1[,1])),breaks = k_cross,labels=FALSE)
  f2<-cut(seq(1,length(dataclass2[,1])),breaks = k_cross,labels=FALSE)
  train.i<-list()
  test.i<-list()
  for(k in 1:k_cross){
    test.i1<-which(f1==k,arr.ind=TRUE)
    test.i2<-which(f2==k,arr.ind=TRUE)
    test.i[[k]]<-row.names(rbind(dataclass1[test.i1,],dataclass2[test.i2,]))
    train.i[[k]]<-row.names(rbind(dataclass1[-test.i1,],dataclass2[-test.i2,]))

  }
  indexTrainTest = list(train.i,test.i)
  return(indexTrainTest)
}