Mining Unique-m Substrings from Genomes