diff plotwithscale.R @ 0:800c7e974e3b draft

planemo upload for repository https://github.com/fabio-cumbo/IWTomics4Galaxy commit 90c984bb9f803e1531fe4cb670e876809d48d205
author iuc
date Fri, 04 Aug 2017 17:51:00 -0400
parents
children
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/plotwithscale.R	Fri Aug 04 17:51:00 2017 -0400
@@ -0,0 +1,117 @@
+if (require("IWTomics",character.only = TRUE,quietly = FALSE)) {
+  args=commandArgs(TRUE)
+  
+  # get args names and values
+  args_values=strsplit(args,'=')
+  args_names=unlist(lapply(args_values,function(arg) arg[1]))
+  names(args_values)=args_names
+  args_values=lapply(args_values,function(arg) arg[2])
+  # read filenames
+  adjustedpvalue=args_values$adjustedpvalue
+  iwtomicsrespdf=args_values$iwtomicsrespdf
+  iwtomicssumpdf=args_values$iwtomicssumpdf
+  iwtomicsrdata=args_values$iwtomicsrdata
+  iwtomicstests=args_values$iwtomicstests
+  iwtomicsselectedfeatures=args_values$iwtomicsselectedfeatures
+  test_subset=paste0('c(',strsplit(args_values$test_subset,'\\|')[[1]],')')
+  feature_subset=paste0('c(',strsplit(args_values$feature_subset,'\\|')[[1]],')')
+  # read parameters (from test_subset on)
+  i_scale_subset=which(args_names=='scale_subset')
+  for(i in i_scale_subset:length(args)){
+    eval(parse(text=args[[i]]))
+  }
+  
+  # load RData
+  load(iwtomicsrdata)
+  # read testids and featureids and check them
+  unlisted=lapply(seq_along(test_subset),
+                   function(i){
+                     test_subset_i=eval(parse(text=test_subset[i]))
+                     feature_subset_i=eval(parse(text=feature_subset[i]))
+                     test_subset_i=rep(test_subset_i,each=length(feature_subset_i))
+                     feature_subset_i=rep(feature_subset_i,length.out=length(test_subset_i))
+                     scale_subset_i=rep(scale_subset[i],length(test_subset_i))
+                     return(list(test_subset=test_subset_i,feature_subset=feature_subset_i,scale_subset=scale_subset_i))
+                   })
+  test_subset=unlist(lapply(unlisted,function(l) l$test_subset))
+  feature_subset=unlist(lapply(unlisted,function(l) l$feature_subset))
+  scale_subset=unlist(lapply(unlisted,function(l) l$scale_subset))
+  testids=as.character(read.delim(iwtomicstests,header=FALSE,sep='\t',stringsAsFactors=FALSE))
+  featureids=as.character(read.delim(iwtomicsselectedfeatures,header=FALSE,sep='\t',stringsAsFactors=FALSE))
+  id_features_subset=featureids[feature_subset]
+  if(sum(testids!=paste(testInput(regionsFeatures_test)$id_region1,'vs',testInput(regionsFeatures_test)$id_region2))){
+    write("Wrong test ids.", stderr())
+    quit(save="no", status=10)
+  }
+  if(sum(featureids!=idFeatures(regionsFeatures_test))){
+    write("Wrong feature ids.", stderr())
+    quit(save="no", status=20)
+  }
+  # retrieve test and features_subset ids
+  id_features_subset=featureids[feature_subset]
+  if(sum(duplicated(paste0(test_subset,id_features_subset)))){
+    write("Two scale thresholds selected for the same test and feature.", stderr())
+    quit(save="no", status=30)
+  }
+  # If scale_subset=0, do not change the threshold
+  default=(scale_subset==0)
+  scale_subset=scale_subset[!default]
+  test_subset=test_subset[!default]
+  id_features_subset=id_features_subset[!default]
+  
+  # get scale threshold
+  scale_threshold=lapply(regionsFeatures_test@test$result,
+                         function(result) unlist(lapply(result,function(feature) feature$max_scale)))
+  for(i in seq_along(test_subset)){
+    if(scale_threshold[[test_subset[i]]][id_features_subset[i]]<scale_subset[i]){
+      write("Scale threshold too high.", stderr())
+      quit(save="no", status=40)
+    }
+    scale_threshold[[test_subset[i]]][id_features_subset[i]]=scale_subset[i]
+  }
+  
+  # create adjustedvalue output
+  pval=adjusted_pval(regionsFeatures_test,scale_threshold=scale_threshold)
+  for(test in seq_along(pval)){
+    for(id_feature in idFeatures(regionsFeatures_test)){
+      write(paste0('Test: ',testids[test],', on feature ',id_feature),
+              file=adjustedpvalue,append=TRUE,sep='\t')
+        pval_i=as.data.frame(t(pval[[test]][[id_feature]]))
+        row.names(pval_i)=paste('Scale',scale_threshold[[test]][[id_feature]])
+        write.table(pval_i,file=adjustedpvalue,append=TRUE,sep='\t',quote=FALSE,row.names=TRUE,col.names=FALSE)
+        write('',file=adjustedpvalue,append=TRUE,sep='\t')
+      }
+    }
+  
+  
+  # plot test results
+  pdf(iwtomicsrespdf,width=5,height=7)
+  if(plottype=='boxplot'){
+    # fix repeated probs
+    probs=sort(unique(probs))
+  }else{
+    probs=c(0.25,0.5,0.75)
+  }
+  plotTest(regionsFeatures_test,alpha=testalpha,type=plottype,probs=probs,average=average,size=size,scale_threshold=scale_threshold,ask=FALSE)
+  dev.off()
+  
+  # plot summary results
+  if(groupby!='none'){
+    tryCatch({
+      pdf(iwtomicssumpdf,width=15,height=10)
+      plotSummary(regionsFeatures_test,alpha=summaryalpha,only_significant=only_significant,groupby=groupby,scale_threshold=scale_threshold,ask=FALSE,append=TRUE)
+      dev.off()
+    }, error = function(err) {
+      if (grepl('selected features with different resolution',err$message)) {
+        write("Group by 'test' but selected features with different resolution.", stderr())
+        quit(save="no", status=50) #error: groupby 'test' but selected features with different resolution.
+      }
+      write("Summary plot error. Please try again.", stderr())
+      quit(save="no", status=60) #error 
+    })
+  }
+  
+}else{
+  write("Missing IWTomics package. Please be sure to have it installed before using this tool.", stderr())
+  quit(save="no", status=255)
+}
\ No newline at end of file