If you can not figure out what some code is doing, you could break it up into smaller, isolated chunks (i.e., subs):

use strict;
use warnings;

my %patts;
my %counts;

my $file = 'patterns.txt';
my $fh;
open $fh, '<', $file or die "can not open file $file: $!";
while (<$fh>) {
    chomp;
    my ($word1, $word2) = split /:/;
    $patts{$word1} = $word2;
}
close $fh;

$file = 'text.txt';
open $fh, '<', $file or die "can not open file $file: $!";
while (<$fh>) {
    chomp;
    for my $patt (keys %patts) {
        $counts{$patt} += count_match($_, $patt);
    }
}
close $fh;

for my $patt (keys %counts) {
    print "pattern $patt occurs $counts{$patt} times\n";
}

sub count_match {
    my ($str, $regex) = @_;
    my @words = split /\s+/, $str;
    my $count = 0;
    for (@words) { $count++ if /\b$regex\b/ }
    return $count;
}

__END__
pattern [Aa]tom[oi] occurs 1 times
pattern [Nn]ucle[oi] occurs 3 times
[download]