The following Perl script removes all digits (0-9) and white spaces including "new line". One can easily modify it to remove letters that don't represent any sequences in a sequence file. To write the output to a file, use perl my_perl.pl infile > outfile.

#!perl

#input name of the file to be read from command line
$infile = @ARGV[0];

#remind use there is no input file
if(!$infile) {
  print "No input file.\nUsage: perl my_perl.pl infile\n";
}


#read the input file if given
if($infile) {
  
  # open inputfile
  open(IN, $infile) or die "can not open $infile\n";

  #read the file
   while( $line =<IN> ) {

     #remove digits and spaces
     $line =~ s/[\d\s]//g;

     print "$line";
   }

  close(IN) or die "can not close $infile\n";

}