examples/A novel algorithmic approach to Bayesian logic regression supplementaries/proceed/postproceed.r

In aliaksah/EMJMCMC2016: EMJMCMC

library(hash)
library(stringi)


#read the output files of all of the MM simulations for the scenario of interest
temp = list.files(pattern="postLog2etaOld_*")
myfiles = lapply(FUN = read.csv,X = temp)

#create a dummy matrix of binary covariates
X= as.data.frame(array(data = rbinom(n = 50*1000,size = 1,prob = runif(n = 50*1000,0,1)),dim = c(1000,50)))
rhash=hash()


#specify true positive discoveries for the scenario of interest
scenairo=1
if(scenairo == 1|2)
{
  tps=array(data = 0, dim =c(1000,3))
  tps[,1]=((1-X$V1)*X$V4)
  tps[,2]=(X$V8*X$V11)
  tps[,3]=(X$V5*X$V9)
  
}

if(scenairo==3)
{
  tps=array(data = 0, dim =c(1000,3))
  tps[,1]= X$V4*X$V17*X$V30*X$V10
  tps[,2]= X$V7*X$V20*X$V12
  tps[,3]= X$V2*X$V9
}

if(scenairo == 4)
{
  tps=array(data = 0, dim =c(1000,3))
  tps[,1]=(X$V1*X$V4)
  tps[,2]=(X$V8*X$V11)
  tps[,3]=(X$V5*X$V9)
}

if(scenairo == 5)
{
  tps=array(data = 0, dim =c(1000,4))
  tps[,1]= X$V4*X$V17*X$V30*X$V10
  tps[,2]= X$V7*X$V20*X$V12
  tps[,3]=(X$V2*X$V9)
  tps[,4]=(X$V37)
}

if(scenairo==6)
{
  tps=array(data = 0, dim =c(1000,8))
  tps[,1]= X$V4*X$V17*X$V30*X$V10
  tps[,2]= X$V37*X$V20*X$V12
  tps[,3]= X$V2*X$V9
  tps[,4]= X$V1*X$V27*X$V3
  tps[,5]= as.integer((X$V50*X$V19+X$V13*X$V11)>0)
  tps[,6]= X$V21*X$V18
  tps[,7]= X$V7
  tps[,8]= X$V8
}

if(scenairo==6.5)
{
  tps=array(data = 0, dim =c(1000,11))
  tps[,1]= X$V4*X$V17*X$V30*X$V10
  tps[,2]= X$V37*X$V20*X$V12
  tps[,3]= X$V2*X$V9
  tps[,4]= X$V1*X$V27*X$V3
  tps[,5]= as.integer((X$V50*X$V19+X$V13*X$V11)>0)
  tps[,6]= X$V21*X$V18
  tps[,7]= X$V7
  tps[,8]= X$V8
  tps[,9]= X$V50*X$V19*X$V13*X$V11
  tps[,10]= X$V50*X$V19
  tps[,11]= X$V13*X$V11
}

#analyse the results of the simulations
N=length(myfiles)
alpha=0.5
clear(rhash)
fdr=0
for(i in 1:min(100,N))
{
  tpi=0
  fpi=0
  for(j in 1:length(myfiles[[i]]$posterior))
  {
    if(myfiles[[i]]$posterior[j]>=alpha)
    {
      expr=as.character(myfiles[[i]]$tree[j])
      print(expr)
      res=model.matrix(data=X,object = as.formula(paste0("V1~",expr)))
      res[,1]=res[,1]-res[,2]
      TP=0
      for(u in 1:dim(tps)[2])
      {
        if(sum(tps[,u]-res[,2])==0)
        {
          TP=1
          break
        }
      }
      if(TP)
      {
        tpi=tpi+1
      }else{
        fpi=fpi+1
      }
      
      ress=c(stri_flatten(res[,1],collapse = ""),stri_flatten(res[,2],collapse = ""),1,expr)
      if(!(ress[1] %in% values(rhash)||(ress[2] %in% values(rhash))))
        rhash[[ress[1]]]=ress
      else
      {
        if(ress[1] %in% keys(rhash))
        {
          rhash[[ress[1]]][3]= (as.numeric(rhash[[ress[1]]][3])) + as.numeric(1)
          if(stri_length(rhash[[ress[1]]][4])>stri_length(expr))
            rhash[[ress[1]]][4]=expr
        }
        else
        {
          rhash[[ress[2]]][3]= (as.numeric(rhash[[ress[2]]][3])) + as.numeric(1)
          if(stri_length(rhash[[ress[2]]][4])>stri_length(expr))
            rhash[[ress[2]]][4]=expr
        }
      }
    }
  }
  print(fpi)
  print(tpi)
  if((fpi+tpi)>0)
    fdr=fdr+(fpi/(fpi+tpi))
}

#print the FDR out
print("FDR:")
print(fdr/100)


#write the frequencies of all significant discoveries
res=as.data.frame(t(values(rhash)[c(4,3),]))
res$V1=as.numeric(as.character(res$V1))
res=res[order(res$V1, decreasing = T),]
colnames(res)=c("posterior","tree")
write.csv(x = t(values(rhash))[,c(3,4)],file = "found.csv",row.names = F,col.names = F)