Figure 1
From: Score-based prediction of genomic islands in prokaryotic genomes using hidden Markov models
States and transition probabilities of SIGI-HMM's Markov chain. The state NATIVE represents genes which are unsuspicious with respect to synonymous codon frequencies. For ρ = 1, 2,..., r, the state PUTALρ models genes, whose codon usage resembles more the prevalences of genomes ρ which represents taxon ρ . Each transition from state x to state y is characterized by its transition probability px 2y. In order to model the mosaic structure of GI composition, transitions from any state PUTAL ρ to any other one PUTAL σ are allowed.