Based On HMM And ANN Gene Identification Method

Bioinformatics is an emerging cross-disciplinary, the use of modern technology on the molecular biology of the data collected for maintenance and management, combined with multi-disciplinary knowledge of these data analysis to the phenomenon of life for a reasonable explanation and prediction. Implicit Ma model is commonly used in biological information model, it first used voice recognition, in the sphere of biological information used in the sequence than right, and fragments of the sequence data mining and classification, structural analysis and pattern recognition.Artificial Neural Networks (ANN) as a simulation of the human brain mechanisms of the engineering model of thinking, and HMM On the contrary, its classification decision-making ability and the ability to describe the uncertainty of information has been universally acknowledged, but its dynamic ability to describe the time signals do not Less than satisfactory, usually ANN classification can only solve the problem of static pattern classification and does not involve processing sequence.Hidden Markov Model is a chronology based on the cumulative probability of dynamic information-processing method. In training and recognition, HMM model parameters from a similar mode of training samples are collected, one for each type of model HMM model, when the need to learn a new model, simply that the model and the corresponding categories of HMM, no Changes in other types of HMM, with a better ability to learn and study again. But the shortcomings of HMM is only considered the characteristics of the type of change, while ignoring the characteristics of the overlap between the categories, only in accordance with the cumulative probability of HMM categories determined to do the maximum, leading to some confusing difficult to identify the gene. HMM and to combine artificial neural network, constructed a new model identification HMM / ANN, it not only overcome the HMM itself difficult to resolve the mode of overlap between the categories, but make up for a neural network to obtain information on the lack of timing.This paper briefly introduced the implicit Ma artificial neural network models and mathematical principles, followed by discussion of the HMM HMM and artificial neural network model ANN in gene identification of specific applications, presented their respective model training and recognition algorithm, Finally, a strong HMM of time to the entire capacity and a strong Category ANN ability to use their respective advantages and the HMM /ANN combine, and Ecogene database of 832 genes as an example, in its identification of genes The application, experiments show that the mixed model of HMM / ANN improve the accuracy of gene identification, fully embodies the mixed model of the feasibility and effectiveness.