R/ga_cox_boost_fun.R

In SydneyBioX/SurvBenchmark_package: survival approaches benchmarking

#' CoxBoost model using GA as feature selection
#'
#'
#' @param r a numeric value, a seed to run this method
#' @param data a dataframe, the data used to performance this survival model
#' @param cvK a numeric value, cross-validation fold
#' @param numm a numeric value, the number of variables,i.e.for example, number of proteins in the data
#' @param topnumm a numeric value, the number of variables selected to be passed into the model, for example, the number of DE genes
#' @param generation_num a numeric value, the generation number used in the GA algorithm, details see package "GenAlgo"
#' @param timess a numeric vector of length 15, contains time points to get the time-dependent AUC values
#' @param time1 a numeric value, the time point to calculate the risk, see package "CoxBoost"
#' @param stepnumber a numeric value, the number of stpes performed in the model, see package "CoxBoost"
#' @param penaltynumber a numeric value, the penalty number used in the model, see package "CoxBoost"
#' @return a data.frame with allevaluation measurements in all columns and rows are each fold results from cross-validation
#'
#' @examples
#' data("exampledt", package = "SurvBenchmark")
#' fitform_ogl=survival::Surv(time,status)~.
#' formula1=fitform_ogl
#' formula2=fitform_ogl
#' formula3=survival::Surv(time,status)~1
#' formula4=survival::Surv(time,status)~1
#' form1=as.formula(~.)
#' timess=seq(as.numeric(summary(cancerdt2_1$time)[2]),as.numeric(summary(cancerdt2_1$time)[5]),(as.numeric(summary(cancerdt2_1$time)[5])-as.numeric(summary(cancerdt2_1$time)[2]))/14)
#' time1=timess[3]
#' stepnumber=10
#' penaltynumber=100
#' want=ga_cox_boost_fun(1,cancerdt2_1,5, 16047,5,20,time1,timess,stepnumber,penaltynumber);
#' @export
#'
ga_cox_boost_fun=function(r,data,cvK,numm,topnumm, generation_num,time1,timess,stepnumber,penaltynumber){
  if (! is.numeric(r)) stop("Input seed is wrong")
  if (! is.numeric(cvK)) stop("Input cross-validation fold number is wrong")
  if (! is.numeric(numm)) stop("Input number of variables is wrong")
  if (! is.numeric(topnumm)) stop("Input number of top variable selection is wrong")
  if (is.null(dim(data))) stop("Input data is wrong")
  if (length(timess)!=15) stop("Wrong time vector length")
  if (class(timess)!= "numeric") stop("Wrong time vector type")
  set.seed(r)
  print(r)
  cvSets = cvTools::cvFolds(nrow(data), cvK)  # permute all the data, into 5 folds
  bicfun=purrr::possibly(function(j){
    test_id = cvSets$subsets[cvSets$which == j]
    test = data[test_id, ]
    train = data[-test_id, ]
    train2=train[!train$os_class=="not",]
    test2=test[!test$os_class=="not",]

    ##ga feature selection
    Data=t(train2[,1:numm]) #gene by patient
    myContext <- list(dataset=Data, gps=train2$os_class)
    n.individuals <- round(numm*0.8,digits = 0) #searching into 80% of genes
    n.features <- topnumm #how many genes we would like to select
    y <- matrix(0, n.individuals, n.features)
    for (i in 1:n.individuals) {

      y[i,] <- sample(1:nrow(Data), n.features)
    }
    mahaFitness <- function(arow, context) {
      GenAlgo::maha(t(context$dataset[arow,]), context$gps, method='var')
    }
    my.ga <- GenAlgo::GenAlg(y, mahaFitness, GenAlgo::selectionMutate, myContext, 0.001, 0.75) #there might be sigularity issues
    for (i in 1:generation_num) {
      my.ga <- GenAlgo::newGeneration(my.ga)
    }
    #summary(my.ga)
    selectedname <- rownames(Data[my.ga@best.individual,])
    #print(selectedname)



    train=train[,colnames(train)%in%c(selectedname,"status","time")]
    test=test[,colnames(test)%in%c(selectedname,"status","time")]


    #fitform_ogl=as.formula(paste("Surv(time, status)~ ", paste(colnames(train)[1:(dim(train)[2]-2)], collapse= "+")))
    fitform_ogl=survival::Surv(time,status)~.
    form1=as.formula(~.)
    formula1=fitform_ogl
    formula2=fitform_ogl
    formula3=survival::Surv(time,status)~1
    formula4=survival::Surv(time,status)~1

    #form1=as.formula(paste("~ ",paste(colnames(train)[1:(dim(train)[2]-2)], collapse= "+")))

    tr_predictormatrix=train[,-which(colnames(train)%in% c("time","status"))]
    tr_predictormatrix=data.matrix(tr_predictormatrix)
    te_predictormatrix=test[,-which(colnames(test)%in% c("time","status"))]
    te_predictormatrix=data.matrix(te_predictormatrix)
    coxboost=CoxBoost::CoxBoost(time=train$time,status=train$status,x=tr_predictormatrix,stepno=stepnumber,penalty=penaltynumber)
    lpnew=predict(coxboost,type="risk",times=time1,newdata=te_predictormatrix)
    #it is the survival probability #the predicted probability of not yet having had the event at the time points given in times
    lpnew=-lpnew #change to risk

    lp<- predict(coxboost,type="risk",times=time1,newdata=tr_predictormatrix)
    lp=-lp
    #cindex
    harrelC1 <- Hmisc::rcorr.cens(-lpnew,survival::Surv(test$time,test$status))
    hc_1<-harrelC1["C Index"]
    Surv.rsp <- survival::Surv(train$time, train$status)
    Surv.rsp.new <- survival::Surv(test$time, test$status)
    bc_1 <- suvAUC::BeggC(Surv.rsp, Surv.rsp.new,lp, lpnew)
    unoc_1<-suvAUC::UnoC(Surv.rsp, Surv.rsp.new, lpnew)
    ghc_1<-suvAUC::GHCI(lpnew)


    #br
    briers1 <- suvAUC::predErr(Surv.rsp, Surv.rsp.new, lp, lpnew,times=test$time, type = "brier", int.type = "unweighted")$error
    br1<-sum(na.omit(briers1))
    briers2<-suvAUC::predErr(Surv.rsp, Surv.rsp.new, lp, lpnew,times=test$time, type = "brier", int.type = "weighted")$error
    br2<-sum(na.omit(briers2))
    ibsfun1=purrr::possibly(function(modell){
      briers3 <- pec::pec(list("cox1"=modell),data=test,formula=fitform_ogl1,cens.model="cox")
      return(crps(briers3)[2])
    },otherwise = NA)
    br3<-NA
    ibsfun2=purrr::possibly(function(modell){
      briers4 <- pec::pec(list("cox1"=modell),data=test,formula=fitform_ogl2,cens.model="marginal")
      return(crps(briers4)[2])
    },otherwise = NA)
    br4<-NA
    ibsfun3=purrr::possibly(function(modell){
      briers5 <- pec::pec(list("cox1"=modell),data=test,formula=fitform_ogl3,cens.model="cox")
      return(crps(briers5)[2])
    },otherwise = NA)
    br5<-NA
    ibsfun4=purrr::possibly(function(modell){
      briers6 <- pec::pec(list("cox1"=modell),data=test,formula=fitform_ogl4,cens.model="marginal")
      return(crps(briers6)[2])
    },otherwise = NA)
    br6<-NA
    #time-dependent auc
    times <- timess
    AUC_CD <- suvAUC::AUC.uno(Surv.rsp, Surv.rsp.new, lpnew, times)
    a1=AUC_CD$auc[1]
    a2=AUC_CD$auc[2]
    a3=AUC_CD$auc[3]
    a4=AUC_CD$auc[4]
    a5=AUC_CD$auc[5]
    a6=AUC_CD$auc[6]
    a7=AUC_CD$auc[7]
    a8=AUC_CD$auc[8]
    a9=AUC_CD$auc[9]
    a10=AUC_CD$auc[10]
    a11=AUC_CD$auc[11]
    a12=AUC_CD$auc[12]
    a13=AUC_CD$auc[13]
    a14=AUC_CD$auc[14]
    a15=AUC_CD$auc[15]
    a=AUC_CD$iauc
    return(c(hc_1,bc_1,unoc_1,ghc_1,br1,br2,br3,br4,br5,br6,a1,a2,a3,a4,a5,a6,a7,a8,a9,a10,a11,a12,a13,a14,a15,a))},otherwise=NA)

  cv5_result=rbind.data.frame(bicfun(1),bicfun(2),bicfun(3),bicfun(4),bicfun(5))
  #colnames(cv5_result)=c("hc_1","bc_1","unoc_1","ghc_1","br1","br2","br3","br4","br5","br6","a1","a2","a3","a4","a5","a6","a7","a8","a9","a10","a11","a12","a13","a14","a15","a")


  return(cv5_result)}