#!/usr/bin/perl

use strict;
use warnings;

chdir "c:/perl/sam" or die "can't chdir to c:/perl/sam: $!";

# get the user-specified sequence file, library file
# and cut-off percentage

my ( $seqfile, $libfile, $cutoff ) = @ARGV;
die "Usage: $0 seq.file lib.file cutoff\n"
    unless ( @ARGV == 3 and -f $seqfile and -f $libfile
             and $cutoff =~ /^\d+\.?\d*$/ );

# run fasta34.exe with the sequence file and library file

my $main = "main.fasta";
my $result = system("fasta34.exe -O $main -Q $seqfile $library");
if ($result >> 8) {
    warn "fasta34.exe ended with non-zero exit status \n";
}

# open the file that was output by fasta34

open(FASTA,"$main") or die "cant open fasta34 output file: $!";

$/ = '>>';  # set INPUT_RECORD_SEPARATOR to ">>"

while (<FASTA>)
{
    chomp;  # this removes ">>" from the end of the string

    next if ( /^\s*$/ );  # skip the first read (update: decided not to test length())

    ($per) = /(\d+\.\d+)% identity/;

    push @org, ">>$_";
    push @input, ">>$_" if ($per > $cutoff);
}
close FASTA;

# if you want to save @input and/or @org to a file and run fasta34 again:

open OUT, ">nextinput.fasta" or die $!;
print OUT join '', @input;
close OUT;

# and likewise for @org