my (%data);
while (<DATA>) {
  chomp;
  # $_ contains '100-233,MA,150:250', for example
  my @initialparts = split( /,/, $_ );
  # @initialparts now contains ( '100-233', 'MA', '150:250' )
  my @leadparts = split( /-/, $initialparts[0]);
  # @leadparts now contains ( '100', '233' )
  push( @{$data{$leadparts[0]}{$initialparts[1]}{'items'}, 
      $leadparts[1] );
  # @{$data{'100'}{'MA'}{'items'} = ( '233' )
  if (scalar(@initialparts) > 2) {
    # Do only if there was a last term
    my @tailparts = split( /:/, $initialparts[2] );
    # @tailparts now contains ( '150', '250' )
    push( @{$data{$leadparts[0]}{$initialparts[1]}{'data'}, 
        @tailparts );
  }
}
open(OUTF, $0 . '.out') or die("Can't open $0.out for output: $!\n");
foreach my $f1 (sort(keys(%data))) {
  # Personally, I prefer data sorted, 
      # although you could leave out the sorts above, and following
  print $f1, "\n";
  foreach my $f2 (sort(keys(%{$data{$f1}}))) {
    print $f2, "\n";
    my $v4 = '|'
    if (defined($data{$f1}{$f2}{'data'} {
      $v4 .= join( ':', sort(@{$data{$f1}{$f2}{'data'}) );
    }
    foreach my $v3 (sort(@{$data{$f1}{$f2}{'items'})) {
      print $v3, $v4, "\n";
    }
  }
}
__DATA__
100-233,MA,150:250
100-344,MA,350:
200-400,ER,
200-300,ER,576
100-250,MA,150
75-300,MA,350