genome_prepare/script/gtf2id_table.pl

#!perl -w
if($ARGV[0]=~/gz$/){
	open GTF, "gzip -dc $ARGV[0]|" or die $!;
}else{
	open GTF, "$ARGV[0]" or die $!;
}

my %allgene;
while (<GTF>) {
    chomp;
    my ($gene_id,$gene_type,$locus_tag,$old_locus_tag,$entry_id,$protein_id)=("-","-","-","-","-","-");
    if(/\tgene\t/){
        ($gene_id)=$_=~/gene_id \"([^"]+)\"/;
        ($gene_type)=$_=~/gene_biotype \"([^"]+)\"/ if ( $_=~/gene_biotype \"/);
        ($locus_tag)=$_=~/ locus_tag \"([^"]+)\"/  if ( $_=~/ locus_tag \"/);
        ($old_locus_tag)=$_=~/ old_locus_tag \"([^"]+)\"/ if ( $_=~/ old_locus_tag \"/);
        ($entry_id)=$_=~/ \"GeneID:(\d+)/ if ( $_=~/ \"GeneID:/);
        my ($gene_name)=$_=~/gene_name \"([^"]+)\"/;
        my ($gene)=$_=~/gene \"([^\"]+)/;
        my ($Name)=$_=~/Name \"([^\"]+)/;
        my $last_name=$gene_name||$Name||$gene||$locus_tag||$old_locus_tag||$gene_id;
        $allgene{$gene_id}{"type"}=$gene_type;
        $allgene{$gene_id}{"name"}=$last_name;
        $allgene{$gene_id}{"locus"}=$locus_tag;
        $allgene{$gene_id}{"old_locus"}=$old_locus_tag;
        $allgene{$gene_id}{"entry"}=$entry_id;
    }
    if(/gene_id \"/){
        ($gene_id)=$_=~/gene_id \"([^"]+)\"/;
        if(/protein_id \"/){
            ($protein_id)=$_=~/protein_id \"([^"]+)\"/;
        }
        $allgene{$gene_id}{"protein"}{$protein_id}=1 if(/gene_id \"/);
    }
}

close GTF;

print "gene_id\tlocus_tag\told_locus_tag\tentry_id\tprotein_id\tgene_name\tgene_biotype\n";

foreach my $g (sort keys %allgene){
    my @proteins = sort keys %{$allgene{$g}{"protein"}};
    if(@proteins > 1){        
        foreach my $p (@proteins){
            print join("\t",$g,$allgene{$g}{"locus"},$allgene{$g}{"old_locus"},$allgene{$g}{"entry"},$p,$allgene{$g}{"name"},$allgene{$g}{"type"})."\n" if($p ne "-");
        }
    }else{
        print join("\t",$g,$allgene{$g}{"locus"},$allgene{$g}{"old_locus"},$allgene{$g}{"entry"},$proteins[0],$allgene{$g}{"name"},$allgene{$g}{"type"})."\n";
    }
}