#!/usr/local/bin/perl
# find_enzymes.pl

# create a lookup table 
# of restriction enzyme sequences
# keyed by restriction enzyme name
%re_lookup = (
          'Eco47III'=> 'AGCGCT',
          'EcoNI'   => 'CCT.....AGG',
          'EcoRI'   => 'GAATTC',
          'EcoRII'  => 'CC[AT]GG',
          'HincII'  => 'GT[TC][AG]AC',
          'HindII'  => 'GT[TC][AG]AC',
          'HindIII' => 'AAGCTT',
          'HinfI'   => 'GA.TC'
);

# get sequence from user
print "Enter sequence:\n";
$userseq = <STDIN>;
chomp $userseq;

# get a list of restriction enzyme names
my @re_names = keys ( %re_lookup );

print "I will check the sequence for all these enzymes: @re_names\n";

# loop through all the restriction enzymes
foreach $re_name ( @re_names ) {

	# get the sequence of the enzyme
	$re_seq = $re_lookup{$re_name};
	
	# check for the presence of the
	# enzyme using the subroutine defined below
	sequence_match( $userseq, $re_name, $re_seq );
}

print "Done!\n";

#----
# define a subroutine for checking dna sequences for
# the presence of a specific sequence
sub sequence_match {

    # this subroutine takes 3 arguments:
    my ( $dnaseq, $re_name, $re_seq ) = @_;
    
    # report which sequence we are checking
    print "checking $dnaseq for $re_seq\n";

    # do test using regular expression match
    if ( $dnaseq =~ /$re_seq/ ) {
       print "found $re_name in sequence\n";
    }   
}