genome_prepare/script/abstract_lncRNA_gtf.pl

#!perl -w
use strict;
use warnings;
use Getopt::Long;
use ScriptManager;

my ($out,$in);
GetOptions(
        "help|h" =>\&USAGE,
        "o:s"=>\$out,
        "i:s"=>\$in,
);
&USAGE unless ($in && $out);
my @lncRNA_type=qw (lncRNA 3prime_overlapping_ncRNA antisense bidirectional_promoter_lncRNA lincRNA macro_lncRNA non_coding processed_transcript sense_intronic sense_overlapping TEC lnc_RNA);
my %type;
map {$type{$_}=1} @lncRNA_type;

open IN,$in or die $!;
open OUT,">$out";
while (<IN>) {
	next if(/^\#/);
	print OUT $_ if(/gene_biotype \"([^\"]+)\"/ && $type{$1});
        print OUT $_ if(/transcript_biotype \"([^\"]+)\"/ && $type{$1});
}
close IN;
close OUT;

# remove $out if $out is empty
#unlink $out if -f $out && -z $out;

sub USAGE{
        my $usage=<<"USAGE";
Usage:perl $0 -i reference.gtf -o lncRNA.gtf
Options:
        -i <file>       reference.gtf containing "gene_biotype" attribute, forced
        -o <file>       output file, forced
        -h|help help

USAGE
        print$usage;
        exit;
}