#!/usr/bin/perl
use Bio::SeqIO;

#=================================
#Script to split fasta file into multiple files containing specified number of sequences 
#briandotoakleyatarsdotusdadotgov
#k.purdy at warwick.ac dot uk
#=================================

my $usage = "\nUsage:  $0 filename\n\n";
$input_file=$ARGV[0] or die $usage;

print "\nYour input file is: $input_file\n

\nHow many seqs would you like in each output file?  ";
chomp($num_seqs = <STDIN>);

#----define and open output file for first time through loop----
$batch_counter=1;
$output_file=">$batch_counter.fas";
$out = Bio::SeqIO->new(-file => $output_file, -format => 'fasta');
$seq_counter=0;
   
#----Read in file to split------
$seq_in  = Bio::SeqIO->new(-file => "$input_file");

#---loop through seqs---
while ($seq_obj = $seq_in->next_seq()) {
$seq_counter++;

	if ($seq_counter<=$num_seqs)  {
	$out->write_seq($seq_obj);
	}

	else {
	#-----reset sequence counter, define and open output files for next set of seqs----
	$batch_counter++;
	$output_file=">$batch_counter.fas";
	$out = Bio::SeqIO->new(-file => $output_file, -format => 'fasta');
	#---write out first sequence to enter else loop, reset seq_counter to 1---
	$out->write_seq($seq_obj);
	$seq_counter=1;	
	}

}


print "\nDone.\n\nInput file split into ",$batch_counter-1, " batches
of $num_seqs each plus remainder in last file\n\n";