faclean

#!/usr/bin/perl
# Reformat and optionally remove short contigs from a FASTA file
# Copyright 2012 Shaun Jackman

use strict;
use Getopt::Std qw'getopts';

my %opt;
getopts 'c:l:L:', \%opt;
my $opt_columns = defined $opt{'c'} ? $opt{'c'} : 0;
my $opt_min_length = defined $opt{'l'} ? $opt{'l'} : 0;
my $opt_max_length = defined $opt{'L'} ? $opt{'L'} : 2000000000;

while (<>) {
	next if /^#/;
	die unless /^>/;
	chomp;
	#my $header = $_;
	my ($id, $comment) = split ' ', $_, 2;

	my $seq = '';
	while (<>) {
		next if /^#/;
		last if /^>/;
		chomp;
		$seq .= $_;
	}

	# Skip short sequences.
	my $len = $seq =~ tr/ACGTacgt//;
	if ($opt_min_length <= $len && $len < $opt_max_length) {
		#print $header, "\n";
		#print $id, ' ', $comment, "\n";
		print $id, "\n";
		if ($opt_columns > 0) {
			print substr($seq, 0, $opt_columns, ''), "\n"
				while length $seq > 0;
		} else {
			print $seq, "\n";
		}
	}

	redo if /^>/;
	last if eof;
}