## Replace gene_id column with gene_name column in the gtf file for RSEM.
## Also check if any transcript_id defined in multiple chromosomes.

my ($fileName) = @ARGV;
if (not defined $fileName) {
    die "File name is required.";
} else {
  print "valid_$fileName and notvalid_$fileName will be created";
}


open(OUT1, ">valid_$fileName");
open(OUT2, ">notvalid_$fileName");
my %transcipt;
my $file = $fileName;
open IN, $file;
while( my $line = <IN>)  {
    chomp;
    @a=split("\t",$line);
    @attr=split(";",$a[8]);
    my %h;
    for my $elem (@attr) {
        ($first, $rest) = split ' ', $elem, 2;
        $h{$first} = $rest.";";
    }
    my $geneId = "";
    my $transcript_id = "";
    if (exists $h{"gene_name"}){
        $geneId = $h{"gene_name"};
    } elsif (exists $h{"gene_id"}){
        $geneId = $h{"gene_id"};
    }
    if (exists $h{"transcript_id"}){
        $transcript_id = $h{"transcript_id"};
    } elsif (exists $h{"transcript_name"}){
        $transcript_id = $h{"transcript_name"};
    } elsif (exists $h{"gene_id"}){
        $transcript_id = $h{"gene_id"};
    }
    if ($geneId ne "" && $transcript_id ne ""){
	## check if any transcript_id defined in multiple chromosomes.
	if (exists $transcipt{$transcript_id}){
             if ($transcipt{$transcript_id} ne $a[0]){
	       print OUT2 "$transcript_id: $transcipt{$transcript_id} vs $a[0]\n";
	        next;
		}
        } else {
             $transcipt{$transcript_id} = $a[0];
	}



	$a[8]=join(" ",("gene_id",$geneId,"transcript_id",$transcript_id));
        print OUT1 join("\t",@a), "\n";
    }  else {
        print OUT2 "$line";
    }
}
close OUT1;
close OUT2;
close IN;