Probabilistic methods of identifying genes in prokaryotic genomes: Connections to the HMM theory