Protein-dna Structural Model Of Transcription Factor Binding Site Prediction,

The prediction of transcription factor binding sites (TFBS) is a hot issue in bioinformatics, its purpose is searching regulatory element on upstream of gene sequence that has critical influence to gene expression & regulation. As transcriptional regulatory elements, TFBS are some kind of DNA sequence fragments, its recognition & prediction is the key of understanding the transcriptional regulatory mechanism and gene expression pattern. With the development of sequence alignment algorithm and computer technology, TFBS recognition under computer platform becomes an important auxiliary research tool. Quick and accurate TFBS identification approach is helpful to biologists for locating target gene of transcription factor on further study of transcription regulation, and it is also provides valuable reference information to molecular biological experiment. Up-to-date, in this area many prediction & recognition software have been developed, such as MEME, AlignACE and BioProspector.This paper puts forward a TFBS prediction method based on Protein-DNA structural model, our method overcomes two problems in the existing classic sequence alignment algorithms:①Identification precision is limited by prior knowledge;②Theoretical model is insufficiency on biological significance. Our method is based on the molecular structural characteristics parameters of Protein-DNA complex three dimensional coordinate file in PDB database. Through the establishment of corresponding chemical, thermal dynamic specific-score position weight matrices, the method uses ensemble approach to predict binding sits and output the TFBS set. We made experiments on Jaspar, which is a transcription factor binding sites profile database, and compare our method with other classic structural prediction algorithms. Results show that our method has good performance and accurate prediction ability to identify TFBS, it is effective and feasible.Here our TFBS prediction algorithm of Protein-DNA structural model uses position weight matrix description to represent various structural characteristics of transcription factors in the identification of binding sites specificity, such measurement can provide important clues to future TF-DNA interaction analysis for structural characteristics and effective quantitative analysis tool for large bio-molecular interaction. To those biologists who hope employ analysis on biological molecular structural characteristics, our work has an important consultant significance.