We developed a computational model to explore the hypothesis that regulatory instructions are context dependent and conveyed through specific 'codes' in human genomic DNA. We provide examples of correlation of computational predictions to reported mapped DNase I hypersensitive segments in the HOXA locus in human chromosome 7. The examples show that statistically significant 9-mers from promoter regions may occur in sequences near and upstream of transcription initiation sites, in intronic regions, and within intergenic regions. Additionally, a subset of 9-mers from coding sequences appears frequently, as clusters, in regulatory regions dispersed in noncoding regions in genomic DNA. The results suggest that the computational model has the potential of decoding regulatory instructions to discover candidate transcription factor binding sites and to discover candidate epigenetic signals that appear in both coding and regulatory regions of genes.