You can use the same code because the \s* and \s+ in the match will also cater for newline (and carriage return if on Windows) so extracting the key and value parts will still work.

$ perl -MData::Dumper -E '
> $text = <<EOT;
> a 2 b 4,5,6 c d e 45,657,-67
> f g 23,54 h 5 i j
> 286,71,90 k 7 l 72
> m 26,8 n
> EOT
> %hash =
>    map { $_->[ 0 ] => [ split m{,}, $_->[ 1 ] ] }
>    sub {
>        push @arr, [ $1, $2 ]
>           while $_[ 0 ] =~ m{\s*([a-z])\s+([-,\d]*)}g;
>        return @arr;
>        }->( $text );
> say Data::Dumper->Dumpxs( [ \ %hash ], [ qw{ *hash } ] );'
%hash = (
          'e' => [
                   '45',
                   '657',
                   '-67'
                 ],
          'n' => [],
          'a' => [
                   '2'
                 ],
          'm' => [
                   '26',
                   '8'
                 ],
          'd' => [],
          'j' => [
                   '286',
                   '71',
                   '90'
                 ],
          'l' => [
                   '72'
                 ],
          'c' => [],
          'k' => [
                   '7'
                 ],
          'h' => [
                   '5'
                 ],
          'b' => [
                   '4',
                   '5',
                   '6'
                 ],
          'g' => [
                   '23',
                   '54'
                 ],
          'f' => [],
          'i' => []
        );
$
[download]

If your data is in a file then slurp the whole of the file into a scalar. Something like (using strictures and warnings):-

my $dataFile = q{/path/to/myDataFile};
my $text = do {
   open my $dataFH, q{<}, $dataFile
      or die qq{open: < $dataFile: $!\n};
   local $/;
   <$dataFH>;
   };
[download]

I hope this answers your question.