#!/usr/bin/perl -w
#######################
#
# Scaffold_Chopper chops up scaffolds into subsets of a user defined length.
# Subsets with "Ns" are disgarded.
#
#
#


use strict;
use warnings;


# Define and initialize Variables

#User Variables

my $scaffold_input = 'scaffold_chopper_sample_input.txt';
my $subset_length = '400';
my $outputfile = 'scaffold_chopper_sample_output.txt';

#Non-User Variables

my $scaffold_record;
my @scaffolds = '';
my $scaffold;
my $scaffold_id;
my $dna;
my $junk1;
my $junk2;
my $current_position;
my $subset_counter;
my $current_subset;

# open input and output files
# input
	unless (open(SCAFFOLD, $scaffold_input)) 

	{
	print "cannot open FASTA total file \"$scaffold_input\"\n\n";
	exit;	
	}

    
#output
	unless (open (SCAFFOLDSUBSETS, ">>$outputfile"))
	{
	print "cannot open file \"$outputfile\" to write to!\n\n";
	exit;
	}
	
	
	

# set input separator to "//\n" and read in a record to a scalar
$/ = "//\n";


# read the data into the scalar called $scaffold_record

$scaffold_record = <SCAFFOLD>;



# separate the file into individual scaffolds

@scaffolds = split />/, $scaffold_record;	


foreach $scaffold (@scaffolds)
	
	{ 
		
		# ensures $scaffold element is actually contains DNA
		if ($scaffold !~ /ATCG/)
			{
			next;
			}
		
		# separate the dna and scaffold id
		
		($junk1, $dna) = split /^.*\n/, $scaffold;
		($scaffold_id, $junk2) = split /$dna/, $scaffold;
	
		# remove the whitespace and newlines from $dna
		
		$scaffold_id =~ s/\n//g;
		$scaffold_id =~ s/\s//g;
	
		$dna =~ s/\n//g;
		$dna =~ s/\s//g;
	
	#	print $dna;
		
		#reset the subset_counter
		
		$subset_counter = 0;
		
		# pull out subsets of user defined length from the scaffold
		
		for ($current_position = 0; $current_position < length $dna; $current_position+= $subset_length) 
			{
			
			
			$current_subset = substr($dna, $current_position, $subset_length);

				#skip over current subsets containing "Ns"
				
				if ($current_subset !~ /N/)
					{
					# print the scaffold_id with the subset counter
					$subset_counter++;
					print SCAFFOLDSUBSETS "\n",">","$scaffold_id","_","$subset_counter\n";
					print_sequence ($current_subset, 60);
					}
			}		

	}

##########Sub-Routines########################################################

# print_sequence (and print_sequence2)
#
# A subroutine to format and print sequence data 

sub print_sequence 

{
    my($sequence, $length) = @_;

    use strict;
    use warnings;

    # Print sequence in lines of $length
    for ( my $pos = 0 ; $pos < length($sequence) ; $pos += $length ) 
    
    {
        print SCAFFOLDSUBSETS substr($sequence, $pos, $length), "\n";
    }
}