Transcription Factor Binding Sites Prediction Algorithm Study And Application

In the post-genome time, the gene research thoroughly causes the research of gene regulatory mechanism, especially transcription regulatory mechanism, to appear urgently. This is because the gene is a foundation of cell's life activity. Each cell of one organism contains the same set of genes, but the expression level of those genes is not the same under different conditions. This kind of behavior decided the RNA composition of cell, then the definition of corresponding proteins, and finally decided the cell function.Transcription not only is the key step that DNA carries on the genetic information shift to the protein a key step, but also is the essential stage that regulates the gene expression. Transcription control is usually realization in the stage of transcription initiation. There are some specific DNA segments in the almost genes upstream region which is called transcription factor binding sites. These segments itself certainly do not carry out any function, only has display the function after it is recognized and bound by transcription factor. Both of them control the gene expression. The union of the transcription factor and corresponding binding sites has the high specificity. The study of transcription factor studies the molecular mechanism of transcription control, studies the union characteristic between a kind of specific protein and the DNA sequence, studies how to regulate transcription by the protein which binds some DNA sequences, and so on. Therefore, if can identify all DNA segments which bind to specific transcription factor, it will be helpful to the study of transcription factor. Detailed experimental work on several individual genes elucidate that the transcription regulatory mechanism in eukaryotes, and especially in higher organisms, is inherently much more complex than moderation of gene expression levels by binding of one single transcription factor to the gene promoter region. In many cases, transcription factors do not work alone, regulation results from the cis-regulatory action of several factors. Therefore, how to identify target sites for cooperatively binding factors is becoming a new luminescent spot.The modern molecular biology experimental techniques to determine whether a given transcription factor which binds any given DNA sequence are well established, such as gel shift analysis and footprinting. However, using these methods to determine whether a given factor binds hundreds or thousands of potential sites is not only very time consuming and labor intensive but also costly. Therefore, the use of computational methods to identify potential transcription factor binding sites is an efficient auxiliary way. But in the actual use process, the over-prediction problem has already seriously restricted these algorithms development.In bioinformatics area, the artificial neural network (ANN) and the hidden Markov model (HMM) have become two importance tools to solve the problem of sequence analysis and model recognition. Here, we reviewed the development course of computational algorithms which search for putative transcription factor binding sites in DNA sequences; discussed the strongpoint and limitation of these methods when the...