On the prediction of genes based on the asymptotic local approach

Authors:
W. K. Yeung;C. W. Chan
Affiliations:
Department of Mechanical Engineering, The University of Hong Kong, Hong Kong, China;Department of Mechanical Engineering, The University of Hong Kong, Hong Kong, China
Venue:
International Journal of Systems Science
Year:
2008

Citing 3
Cited 0

Linear stochastic systems

Linear stochastic systems
Early warning of slight changes in systems

Automatica (Journal of IFAC) - Special issue on statistical signal processing and control
An Intelligent System for Vertebrate Promoter Recognition

IEEE Intelligent Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

A novel gene prediction method based on the detection of the Transcriptional Start Sites of CpG rich genes is presented in this article. By transforming the DNA data into a time series, the detection of the CpG rich region can be reformulated as one that detects a change in the mean of a stochastic process. A statistical test for detecting this change in mean is derived based on the asymptotic local approach. To improve the performance of this method, some properties of the CpG islands approach, such as the cyclic feature, are used to reduce the number of false predictions. The proposed method is applied first to predict genes in the rabbit alpha-like gene cluster. It is shown that all confirmed genes in that sequence are successfully predicted. It is applied next to a section of the human chromosome 22. The proposed method is able to predict 73% of the confirmed genes. To illustrate the performance of the proposed method, a comparison with the Dragon Gene Start Finder method is also made.