CpG islands are stretches of CG-rich sequences in the genome

CpG islands are stretches of CG-rich sequences in the genome. They are often of functional important, as 50% of the human genes have a CpG island around 500bp upstream of the transcription site. Of course, CpG sequence is not only CG, and non-CpG island sequences could still contain some CG. Therefore, we could use HMM to predict CpG islands by looking at a long stretch of DNA. Now as a HMM practise, we just have a short-sequence AGGCGT, can you use HMM to predict the possibility of each base is part of a CpG island sequence? The parameters of the HMM Model are seen in the attachment as well as the probabilities.