الفهرس 
Chapter 1 IntroductionOnly 14 pages are availabe for public view
 |  | from | 162 |  |  |
| | | from | 162 | | |
Abstract
This thesis concerns the identification or detection of protein-coding regions in Deoxyribonucleic Acid (DNA) sequence. Many digital signal processing (DSP) techniques have been applied for identification task. A number of numerical DNA sequence representation have been evolved in order to transform the DNA sequence analysis problems from the traditional string processing domain to the discrete signal processing domain. These methods are electron-ion interaction pseudopotential (EIIP), genetic code context (GCC), 2-bit binary representation and atomic number representation. These methods are used to convert the DNA sequence to numerical values through a mapping function. In addition, a mathematical modeling approach is presented to create closed formulas for the represented data sequences using different methods. The importance of this process is that these mathematical models can be used for any further processing or identification, when applied to DNA sequences. The metrics used for evaluating the accuracy of the processing are Root Mean Square Error (RMSE) and R-square metrics. This thesis presents non-parametric spectral estimation techniques based on the Discrete Fourier Transform (DFT) for the analysis of DNA sequences. These techniques are efficient frequency-domain signal representation techniques, which improve the analysis of DNA sequences and enable the extraction of some desirable information that cannot be extracted from the time-domain representation of these sequences. The adopted techniques are the periodogram, average periodogram (Bartlett), modified average periodogram (Welch), and
Abstract
v
Blackman and Tukey spectral estimation techniques. The objective of these spectral estimation techniques is to investigate the locations of exons in DNA sequences for gene prediction. A comparison study is presented in this thesis between all the suggested spectral estimation techniques from the exon prediction perspective. In addition, non-parametric spectral estimation methods are used for estimating the spectral of both original (actual) and mathematical modeled DNA sequences. Exon detection results based on real and modeled DNA sequences coincide to a great extent, which ensures the success of the proposed sum-of-sinusoids method for modeling of DNA sequences as the best DNA representation mathematical model. This thesis presents parametric spectral estimation methods for the analysis of DNA sequences. These techniques are efficient, which improve the analysis of DNA sequences and enable the extraction of some desirable information of DNA sequences. The adopted techniques are Burg, Covariance, Modified Covariance, Yule Walker, Multiple Signal Classification (MUSIC) and Autoregressive Moving Average (ARMA). The objective of these spectral estimation techniques is to investigate the locations of exons in DNA sequences for gene prediction from DNA sequences. A comparison study is presented in this thesis between these parametric spectral estimation techniques.