Splice Sites Detection by Combining Markov and Hidden Markov Model

Splice sites detection is helpful to the analysis of gene structure and contributes to the prediction of gene products, so it is one of the most important topics in bioinformatics. With the rapid increase of biological data, faster splice sites detecting methods are more appreciated. Markov model and hidden Markov model not only are low-time-cost, but also easy to understand. Besides the detecting result, we can often obtain some biological features from the models, so they are widely used in DNA sites detection. However, the detecting accuracy of the conventional Markov and hidden Markov model is not satisfactory, so in order to improve their detecting accuracy, we propose a dinucleotide-based hidden Markov model, and then combine the Markov model with the dinucleotide-based hidden Markov model to construct an ensemble model. The experiment results show that our models can further improve the detecting accuracy than those of the conventional models and some present models.