--- a/summarize_gff.R	Wed Aug 14 11:24:42 2019 -0400
+++ b/summarize_gff.R	Thu Aug 15 07:20:12 2019 -0400
@@ -1,7 +1,11 @@
 ## summarize hits
 output = commandArgs(T)[2] ## output table
 filepath = commandArgs(T)[1]  ## input dante gff3
-summarized_by = commandArgs(T)[-(1:2)]
+if (length(commandArgs(T))==2){
+  summarized_by = NA
+}else{
+  summarized_by = strsplit(commandArgs(T)[-(1:2)], split = ",")[[1]]
+}
 
 readGFF3fromDante = function(filepath){
   dfraw=read.table(filepath, as.is = TRUE)
@@ -9,29 +13,49 @@
   colnames(gff_df) = c("seqid", "source", "type", "start", "end", "score",
                     "strand", "phase")
   ## assume same order, same attributes names
-  gffattr = do.call(rbind,
-                    lapply(
-                      strsplit(dfraw[,9],split=c("=|;")),
-                      function(x)x[c(FALSE,TRUE)]
-                    )
-                    )
-  gff_df$Name = gffattr[,1]
-  gff_df$Final_Classification = gffattr[,2]
-  gff_df$Region_Hits_Classifications = gffattr[,3]
-  gff_df$Best_Hit = gffattr[,4]
-  gff_df$Best_Hit_DB_Pos = gffattr[,5]
-  gff_df$DB_Seq = gffattr[,6]
-  gff_df$Query_Seq = gffattr[,7]
-  gff_df$Identity = as.numeric(gffattr[,8])
-  gff_df$Similarity = as.numeric(gffattr[,9])
-  gff_df$Relat_Length = as.numeric(gffattr[,10])
-  gff_df$Relat_Interruptions = as.numeric(gffattr[,11])
-  gff_df$Hit_to_DB_Length = as.numeric(gffattr[,12])
+  ## TODO make ti more robust - order can change!
+  gffattr_list = lapply(
+    strsplit(dfraw[,9],split=c("=|;")),
+    function(x)x[c(FALSE,TRUE)]
+  )
+  ## some rows are not complete - in case of ambiguous domains
+  L = sapply(gffattr_list, length)
+  short = L  < max(L)
+  if (any(short)){
+    gffattr_list[short] = lapply(gffattr_list[short],function(x) c(x, rep(NA, 13 - length(x))))
+  }
+  gffattr = as.data.frame(do.call(rbind, gffattr_list))
+
+  ## get attributes names
+  attrnames =  strsplit(dfraw[1,9],split=c("=|;"))[[1]][c(TRUE,FALSE)]
+  colnames(gffattr) = attrnames
+
+  gff_df$Final_Classification = gffattr$Final_Classification
+  gff_df$Name = gffattr$Name
+  gff_df$Region_Hits_Classifications = gffattr$Region_Hits_Classifications
+  gff_df$Best_Hit = gffattr$Best_Hit
+  gff_df$Best_Hit_DB_Pos = gffattr$Best_Hir_DB_Pos
+  gff_df$DB_Seq = gffattr$DB_Seq
+  gff_df$Query_Seq = gffattr$Query_Seq
+  gff_df$Region_Seq = gffattr$Region_Seq
+  gff_df$Identity = as.numeric(gffattr$Identity)
+  gff_df$Similarity = as.numeric(gffattr$Similarity)
+  gff_df$Relat_Length = as.numeric(gffattr$Relat_Length)
+  gff_df$Relat_Interruptions = as.numeric(gffattr$Relat_Interruptions)
+  gff_df$Hit_to_DB_Length = as.numeric(gffattr$Hit_to_DB_Length)
   return(gff_df)
 }
 
 gff = readGFF3fromDante(filepath)
-# summarize_by = c("Final_Classification", "Name", "seqid")
+# summarized_by = c("Final_Classification", "Name", "seqid")
+# summarized_by = c("Final_Classification")
+
 
-tbl = data.frame(table(gff[,summarize_by]))
-write.table(tbl, file = filepath, row.names =FALSE, quote = FALSE)
+if (is.na(summarized_by)){
+  ## export complete table
+  write.table(gff, file = output, row.names = FALSE, quote = FALSE, sep = "\t")
+}else{
+  ## export summary
+  tbl = data.frame(table(gff[, summarized_by]))
+  write.table(tbl, file = output, row.names = FALSE, quote = FALSE, sep = "\t")
+}