ERROR: Objects of type 'close' cannot be partially substituted

>source("text.txt")
>exdata=read.csv("final_output.csv", header=TRUE)
>pred.ML (Y-X1+X2+X4, data=exdata, B=1000) 

pred.ML<-function(formula, data, B=1000, alpha=0.05, Cores=detectCores()){
  library(snow)
  library (foreach)
  library(doParallel)
  library (doSNOW)
  library (pROC)
  library (MASS)

  N<-dim(data)[1]

  boot<-function(formula,data){
  
    N<-dim(data)[1]
    
    bs.i<-sample(1:N,N,replace=TRUE)
    train.i<-data [bs.i,]
    test.i<-data [-bs.i,]

    gm1.i<-lda(formula, data=train.i, family=binomial)
  
    prob.b<-predict(gm1.i, type=c("response")))
    prob.o<-predict(gm1.i, newdata=data, type=c("response")))
    prob.t<-predict(gm1.i, newdata=test.i, type=c("response")))
  
    AUC.b<-roc(train.i$Y to prob.b, levels=c(0,1), direction="<")$auc
    AUC.o<-roc(data$Y to prob.o, levels=c(0,1), direction="<")$auc
    AUC.t<-roc(test.i$Y to prob.t, levels=c(0,1), direction="<")$auc

    return(c(AUC.b, AUC.o, AUC.t))

  }

  gm1<-lda(formula, data=data, family=binomial)

  # calculationapparent AUC&test AUC
  prob<-predict(gm1,type=c("response")))
  ROC.app<-roc(data$Y to prob, levels=c(0,1), direction="<")
  AUC.app<-ROC.app$auc
  delong1<-ci.auc(ROC.app,conf.level=1-alpha)[1]
  delong2<-ci.auc(ROC.app,conf.level=1-alpha)[3]
  
  cl<-makeSOCKcluster (Cores)
  registerDoSNOW(cl)

  block<-ceiling (B/Cores)
  block0<-c(1, block*(1:(Cores-1)+1)
  block1<-c (block*(1:(Cores-1)), B)

  boot.res<-foreach(b=1:Cores, .combine=rbind, .packages=c("MASS", "pROC")%dopar%{

   R1<- NULL
   for(iter in block0[b]:block1[b])R1<-rbind(R1,boot(model,data))
   R1

  }

  # bootstrap SD and 95% CI of parent C
  boot.app.C<-boot.res[,1]
    
  # bias corrected AUC estimate (ordinary bootstrap)
  AUC.boot<-AUC.app-mean(boot.res[,1]-boot.res[,2])   

  AUC.boot.CL1<-quantile (boot.app.C, 0.5*alpha)
  AUC.boot.CL2<-quantile (boot.app.C, 1-0.5*alpha)
    
  # bias corrected AUC estimate (bootstrap.632)
  AUC.loocv<-mean(boot.res[,3])   
  AUC.632<-0.368*AUC.app+0.632*AUC.loocv
  AUC.loocv.SD<-sd(boot.res[,3])
    
  # bias corrected AUC estimate (bootstrap.632+)
  if(AUC.loocv<=0.5){
    R<-1
  } else if(AUC.app>AUC.loocv){
    R<-(AUC.app-AUC.loocv)/(AUC.app-0.5)
  } else{
    R<-0
  }
    
  w<-0.632/(1-0.368*R)
  AUC.632p<-(1-w)*AUC.app+w*max (AUC.loocv, 0.5) 
  
  delta1<-AUC.boot-AUC.app
  delta2<- AUC.632-AUC.app
  delta3<-AUC.632p-AUC.app
  
  R<-list(
   lda.output = gm1,
   N.obs = N,
   N.boot=B,
   C.Apparent=as.numeric (AUC.app),
   C.DeLongCI=c(delong1,delong2),
   C.Apparent_BootstrapCI=c(AUC.boot.CL1, AUC.boot.CL2),
   C. Harrell = AUC.boot,
   C.Harrell_LSCI=(c(AUC.boot.CL1,AUC.boot.CL2) + delta1),
   C.0.632 = AUC.632,
   C.0.632_LSCI=(c(AUC.boot.CL1, AUC.boot.CL2) + delta2),
   C.0.632p = AUC.632p,
   C.0.632p_LSCI=(c(AUC.boot.CL1,AUC.boot.CL2)+ delta3)
   )
   
  stopCluster(cl)
  
  return(R)

}