Perl Hashes and regex - perl

Question

I am working on a code that splits sentence into individual words, the words are then searched against hash keys for their presence. My code returns terms that are 100% same, after a match I tag the word from the sentence with the value that corresponds to the matching key. The problem is the code tags terms but with random values not with what I expect. Also, there are situations where the term and the hash key are similar but not 100% identical,
how can I write a regular expression to match my terms with the keys.
Note: I have stemmed the hash keys to their root forms.
I cam provide some examples: If the term from the sentence is Synergistic or anti-synergistic, and my hash key is Synerg, how can I match the above term with Synerg.
My code is as follows:
open IN, "C:\\Users\\Desktop\\TM\\clean_cells.txt" or die "import file absent";
my %hash=();
use Tie::IxHash;
tie %hash => "Tie::IxHash";
while(<IN>)
{
chomp $_;
$line=lc $_;
#Organs=split/\t/, $line;
$hash{$Organs[0]}=$Organs[1];
}
$Sentence="Lymphoma is Lymph Heart and Lung";
#list=split/ /,$Sentence;
#array=();
foreach $term(#list)
{
chomp $term;
for $keys(keys %hash)
{
if($hash{$term})
{
$cell="<$hash{$keys}>$term<\/$hash{$keys}>";
push(#array, $cell);
}
elsif($term=~m/\b\Q$keys(\w+)\E\b/)
{
$cell="<$hash{$keys}>$term<\/$hash{$keys}>";
push(#array, $cell);
}
elsif($term=~m/\b\Q(\w+)$keys\E\b/)
{
$cell="<$hash{$keys}>$term<\/$hash{$keys}>";
push(#array, $cell);
}
elsif($term=~m/\b\Q(\w+)$keys(\w+)\E\b/)
{
$cell="<$hash{$keys}>$term<\/$hash{$keys}>";
push(#array, $cell);
}
}
}
print #array;
for example: hash looks like this: %hash={
TF1 => Lymph
Thoracic_duct => Lymph
SK-MEL-1 => Lymph
Brain => Brain
Cerebellum => Brain
};
So if the term TF1 is found it should be substituted to Lymph TF1 /Lymph

Answer 1

I found two big problems that were preventing your code from working:
You are making the keys to your hash lowercase, but you are not doing
the same for the terms in $Sentence. Thus, uppercase words from
$Sentence will never match.
The \Q...\E modifier disables regex meta-characters. While it is often good to do this when interpolating a variable, you cannot use expressions like (\w+) in there--that will look for the literal characters (\w+). Those regexes need to be rewritten like this: m/\b\Q$keys\E(\w+)\b/.
There are other design issues with your code, as well:
You are using undeclared global variables all over the place. You should declare all variables with my. Always turn on use strict; use warnings;, which will force you to do this correctly.
There doesn't appear to be any reason for Tie::IxHash, which causes your hash to be ordered. You don't use this ordering in any way in your code. The output is ordered by #list. I would do away with this unnecessary module.
Your if/elsif statements are redundant. if($term=~m/\b\Q(\w*)$keys(\w*)\E\b/) will accomplish the same thing as all of them combined. Note that I replaced \w+ with \w*. This allows the groups before and after to match zero or more characters instead of one or more characters.
Note: I didn't bother testing with Tie::IxHash, since I don't have that module and it appears unnecessary. It's possible that using this module is also introducing other problems in your code.