“…Therefore, developers have to generate their non-promoter sequences. Several strategies for generating non-promoter sequences have been used including: randomly generated sequences [ 16 , 17 , 28 ]; sequences extracted from intergenic or coding regions [ 7 , 11 , 12 , 14 , 15 , 18 , 25 , 28 ]; ii) Feature extraction: Several sequence and structure-based feature representations have been used for developing prokaryotic promoter region prediction methods. Examples of sequence based features include: k-mer representation [ 7 , 12 , 28 , 32 ], variable-window Z-curve [ 18 ], and nucleotide identity (NID) [ 17 ].…”