Perhaps, but if we really want to produce a list of all these numbers, then it comes down to balancing time and space requirements. My original estimation was based on using 40 bytes per sequence and requiring about 10 petabytes. Reducing the storage estimation from 40 bytes to 18 bytes using Billy Conner's suggestion reduces the storage cost to about 5 petabytes. That's nothing to sneeze at. Furthermore, the conversion data formats for a single sequence is O(n), where n is the number of elements within a sequence (in this case 12). Happily the fastest method of producing a sequence would also be O(n), which means algorithmically, Billy Connor's method does not increase the complexity at all while reducing storage by 50%. What's not to love about it?