Incorporating sequence information into the scoring function: a hidden Markov model for improved peptide identification