#!/usr/bin/perl
my $in = ();
my $a = 0;
my $g = 0;
my $r = 0;
for(my $k = $#ARGV; $k >= 0; $k--){
	if($ARGV[$k] eq '-a'){
		$a = 1;
		}
	if($ARGV[$k] eq '-g'){
		$g = 1;
		}
	if($ARGV[$k] eq '-i'){
		if(-e $ARGV[$k+1]){
			$in = $ARGV[$k+1];
			}
		}
	if($ARGV[$k] eq '-r'){
		$r = 1;
		}
	}
if(length($in)){
	open(INFILE, '<', $in) or die("could not open $in!\n");
	my $buffer = ();
	my $name = ();
	my $sequence = ();
	while(my $line = <INFILE>){
		chomp($line);
		if($line =~ m/^>/){
			if(length($sequence)){
				$buffer .= process($name, $sequence);
				if(length($buffer) > 10000){
					print("$buffer");
					$buffer = ();
					}
				}
			$sequence = ();
			$name = $line;
			} else {
				$sequence .= $line;
				}
		}
	$buffer .= process($name, $sequence);
	print("$buffer");
	close(INFILE);
	sub process {
		my @labels = split(/\|/, $_[0]);
		my $output = '>';
		if($a == 1){
			$output .= $labels[3];
			} else {
				my @bits = split(/ /, $labels[4]);
				$output .= join('_', (@bits[1,2], $labels[3]));
				}
		my $sequence = uc($_[1]);
		$sequence =~ tr/ACGTNVDBHWMRKSY//cd;
		if($g == 1){
			my $GC += ($sequence =~ tr/CGNVDBHMRKSY/CGNVDBHMRKSY/);
			$GC = 100*($GC/(1+length($sequence)));
			$output .= '|GC=' . sprintf("%.2f", $GC) . '%';
			}
		if($r == 1){
			$sequence =~ tr/ACGTNVDBHWMRKSY/TGCANBHVDWKYMSR/;
			$sequence = reverse($sequence);
			}
		my @bases = split(//, $sequence);
		for(my $k = 0; $k <= $#bases; $k++){
			if(($k % 80) == 0){
				$output .= "\n" . $bases[$k];
				} else {
					$output .= $bases[$k];
					}
			}
		$output .= "\n";
		return($output);
		}
	} else {
		print("\nGenBank2fasta a script for processing GenBank files.\n\n");
		print("USAGE: GenBank2fasta -i file [-a] [-g] [-r]\n");
		print("-a\treduces names to GenBank accessions only\n");
		print("-g\tappends percent GC content to each name\n");
		print("-i\tspecifies a GenBank fasta file (required)\n");
		print("-r\tproduces the reverse complement of each sequence\n\n");
		}
exit(0);