Just to give you an idea of how little effort it should take to do the thing that your script does, here's a version that:

• allows the user to specify all inputs and options at the command line, so the script knows what to do based on command-line args;
• uses a hash instead of a set of separate scalar variables;
• uses loops for input and output.

#!/usr/bin/perl

use strict;                # enforce variable declarations
use warnings;              # enable warnings

my $Usage = "Usage: $0 [sequence.file]\n  (reads stdin if no file name
+ is given)\n";
die $Usage unless ( -p STDIN or ( @ARGV and -f $ARGV[0] ));

my %counts;  # tally of letter occurrences

# read lines from named file(s) or from stdin:

while (<>) {
    s/\s+$//;  # remove white space (line termination) from end of lin
+e
    for my $ltr ( split // ) {  # split into individual characters
        $counts{$ltr}++;
    }
}

# print results:

for my $ltr ( sort keys %counts ) {
    print " Number of $ltr nucleotides:  $counts{$ltr}\n";
}
[download]

Note that this version will count any character inventory you give it; if the input happens to contain something other than ACGT, you'll get the quantity of occurrence for all the letters that occur in the data.