I'm unsure what you're trying to achieve here, but as I understand it you're taking an input sequence:

  "A C D"
[download]

turning it into a regexp:

  /A(.*?)C(.*?)D/
[download]

matching that against an input sequence:

  "ABICID"
[download]

and want to replace all the characters that didn't directly match with a hyphen:

  "A--C-D"
[download]

If so, one possible approach is to take advantage of the special match arrays @- and @+ - for example for the capture variable $3, $-[3] will be the offset of the start of that capture, and $+[3] the offset just after the end of that capture.

So you could use something like:

  if ($string =~ $test) {
    for my $offset (1 .. $#-) {
      my $start = $-[$offset];
      my $len = $+[$offset] - $start;
      substr($test, $start, $len) = "-" x $len;
    }
    # now extract only the matched extent
    push @hyph_padded_seq, substr($test, $-[0], $+[0] - $-[0]);
  }
[download]

So for the given hash, "A C D" would give results of "A-CD" for key S1, and "A--C-D" for key S4. That doesn't match the "intended answers" you show, but that's the closest I can get to guessing what you're looking for.

Hugo