root/bin/gff3_2_gtf

#! /usr/bin/perl -w
use strict;

my $file = shift;

#parse file
open (IN, "< $file");
my @genes;
my @mRNAs;
my @stuff;

while (defined(my $line = <IN>)){
    #skip comments and fasta
    last if($line =~ /^\#\#FASTA/);
    next if($line =~ /^\#/);

    #add contig lines to header
    my $f = parse($line);
    next if($f->{type} eq 'contig');

    push(@genes, $f) if($f->{type} eq 'gene');
    push(@mRNAs, $f) if($f->{type} eq 'mRNA');
    push(@stuff, $f) if($f->{type} eq 'CDS');
    push(@stuff, $f) if($f->{type} eq 'five_prime_UTR');
    push(@stuff, $f) if($f->{type} eq 'three_prime_UTR');
    push(@stuff, $f) if($f->{type} eq 'start_codon');
    push(@stuff, $f) if($f->{type} eq 'stop_codon');
}
close(IN);

#organize transcript features
my %stuffinx;
foreach my $f (@stuff){
    my $parent = $f->{Parent};
    push (@{$stuffinx{$parent}}, $f);
}

#organize mRNAs
my %mRNAinx;
foreach my $f (@mRNAs){
    my $parent = $f->{Parent};
    push (@{$mRNAinx{$parent}}, $f);
}

foreach my $g (@genes){
    my $gene_id = $g->{ID};
    my $g_mRNAs = $mRNAinx{$g->{ID}};

    foreach my $t (@$g_mRNAs){
        my $transcript_id = $t->{ID};
        my $t_stuff = $stuffinx{$t->{ID}};

        my @UTRs   = grep {$_->{type} =~ /UTR$/} @$t_stuff;
        my @CDSs   = grep {$_->{type} eq 'CDS'} @$t_stuff;
        my @stops  = grep {$_->{type} eq 'stop_codon'} @$t_stuff;
        my @starts = grep {$_->{type} eq 'start_codon'} @$t_stuff;

        @CDSs = fix_CDS(\@CDSs,\@stops);
        
        print_it($gene_id, $transcript_id, @$t_stuff);
    }
}

#------------------------SUBS------------------------
sub fix_CDS {
    my $CDSs  = shift;
    my $stops = shift;

    my @keepers;
    foreach my $CDS (@$CDSs){
        my $cB = $CDS->{start};
        my $cE = $CDS->{end};

        foreach my $stop (@$stops){
            my $sB = $stop->{start};
            my $sE = $stop->{end};

            if($cB == $sB){
                $CDS->{start} += 3;
            }
            elsif($cE == $sE){
                $CDS->{end} -= 3;
            }
        }
        
        push(@keepers, $CDS);
    }

    return @keepers
}

sub print_it {
    my $gene_id = shift;
    my $transcript_id = shift;
    
    ($transcript_id) = $transcript_id =~ /^([^\s\t\n]+)/; 
    while(my $f = shift @_){
        $f->{type} = '5UTR' if($f->{type} eq 'five_prime_UTR');
        $f->{type} = '3UTR' if($f->{type} eq 'three_prime_UTR');
        
        print $f->{seqid} . "\t";
        print $f->{source} . "\t";
        print $f->{type} . "\t";
        print $f->{start} . "\t";
        print $f->{end} . "\t";
        print $f->{score} . "\t";
        print $f->{strand} . "\t";
        print $f->{phase} . "\t";
        print "gene_id \"".$gene_id."\"\;";
        print " transcript_id \"".$transcript_id."\"\;";
        print "\n";
    }
}

sub parse {
    my $l = shift;
    
    chomp $l;
    my @F = split("\t", $l);
    my @att = attrib($F[8]);
    
    my $f = {seqid => $F[0],
             source => $F[1],
             type => $F[2],
             start => $F[3],
             end => $F[4],
             score => $F[5],
             strand => $F[6],
             phase => $F[7],
             ID => $att[0],
             Name => $att[1],
             Parent => $att[2]};
    
    return $f;
}

sub attrib {
    my $att = shift;

    my ($ID) = $att =~ /ID\=([^\;\n]+)/;
    my ($name) = $att =~ /Name\=([^\;\n]+)/;
    my ($parent) = $att =~ /Parent\=([^\;\n]+)/;    
    
    $ID = $name if(! defined $ID);
    $name = $ID if(! defined $name);
    
    return ($ID, $name, $parent);
}