In the Splicing Site Recognition System, a Hidden Markov Model is developed and trained to identify the splicing sites of a DNA sequence. The model is trained using maximum likelihood estimation. A test accuracy of 68.8% is obtained on a dataset consisting of 570 vertebrate gene. Then the state paths are generated using Viterbi algorithm. In this project, Kevin Murphy's HMM toolbox is used to generate the state path. More detail about this project can be found in this link: http://www.cs.ucr.edu/~stelo/cs234winter17/A2.htm. Morever, the implementation details can be found in the "Project Presentation.pdf" file. In order to design the Hidden Markov Model, training and testing process for this problem, I followed a research paper titled "Effective hidden Markov models for detecting splicing junction sites in DNA sequences". This paper can be found in this link: http://www.sciencedirect.com/science/article/pii/S0020025501001608. This project is still under development. A detailed overview of the implementation will be updated gradually.
The sourcecode does not come with any kind of warrenty or support, though you can contact me if you need. Moreover, the project is still under development.
Sharmistha Bardhan