Die Body Similarity Comparison Algorithm Research

Transcription factor (TF) proteins usually recognize a small number of DNA targets via the formation of sequence specific and nonspecific molecular interactions. Understanding the evolution of TF DNA-binding preferences will not only provide useful insights on the mechanism of DNA recognition, it will also allow more accurate prediction of genomic regulatory elements, which still constitutes a major hurdle in understanding cellular gene regulatory networks. Comparative analysis of predicted transcription factor binding site motifs is an important branch of bioinformatics. It is a measure of the effect of the transcription factor binding site recognition algorithm. In this thesis, a similarity comparison method of transcription factor binding site motifs is used to optimize the motif database.We first introduce bioinformatics research; specially focus on the situation and development direction of transcription factor binding site motifs similarity.Then we introduce the concepts of transcription factors and TF binding sites, and analyze transcription factor binding sites representation model. We proposed a new motif similarity comparison algorithm based on a pseudo-band count on the position frequency matrix. The accuracy of our algorithm is better than other existing algorithms through experimental analysis, and using it, we can cluster related motif together, showing the better biological significance. We also parallelize the algorithm, which greatly improve the running speed. This will be very useful when calculating similarity score of a large number of motifs.Finally, we summarize some problems and further research topics of this field.