ncbi的genome，gene序列转换和gb2gtf——链特异性转录组

ncbi下载的数据不一定符合链特异性转录组的格式，写个小脚本进行处理一下：

#!perl 
use warnings;
use strict;

die qq/
perl $0 <type> <outprefix> <file1> [<file2> ...] 

type: "gtf", "gene" and "genome".
Note: "genome" need gtf.list of "gtf", and gtf.list must be the end option of <file...>.

/ if(@ARGV < 3);

my $t = shift @ARGV;
my $out = shift @ARGV;
my $py = "gbk2gtf.py";

if($t eq "gtf")
{
	my @gtf;
	foreach(@ARGV)
	{
		system "cat $_ | python $py > $_.tmp";
		push @gtf, "$_.tmp";
	}

	system "cat @gtf > $out.gtf";
	system "cut -f 1 $out.gtf | sort | uniq > gtf.list";
	system "rm *.tmp -rf";

}elsif($t eq "gene"){
	my @gene;
	foreach my $f(@ARGV)
	{
		open FF, $f or die $!;
		open OUT, "> $f.tmp" or die $!;
		while(<FF>)
		{
			next if(/^\n/);
			chomp;
			if(/^>/)
			{
				my ($id) = $_ =~ /locus_tag=(\w+)/;
				print OUT ">$1\n";
			}else{
				print OUT "$_\n";
			}
		}
		close FF;
		close OUT;
		push @gene, "$f.tmp";
	}

	system "cat @gene > $out.gene.fa";
	system "rm *.tmp -rf";

}elsi