Research On Promoter Recognition Algorithm Based On Support Vector Machine

With the continuous development of genome sequencing,the accumulation speed of related data is increasing exponentially.Research on gene expression has always been the core of genetic research,and promoters play an important role in gene expression.Thus,the study of promoter is very important in genetics.As its name implies,promoters determine the timing of transcription in gene expression,and also influence the intensity of gene expression.The research results at the present stage of promoters have greatly promoted the research of metabolic regulation,construction of expression system and other fields.At present,the main stream promoter prediction algorithms are based on signal characteristics,structure based features and text based feature methods,all of which have their limitations,such as signal features TATA-box,CAAT-box and CpG Island,which are not the only information to determine the function of the promoter,so the prediction success rate of the predictor is not high.The focus of this study is to try to overcome the disadvantages of the traditional promoter prediction methods.By mining the local and global features of the sequence,after comparing the features of a series of machine learning algorithms,a two-layer promoter pretester based on the support vector mechanism is built,and the first layer of predictor is the predictor.If the promoter sequence is not the promoter sequence,and then predict the type of the promoter,that is,whether it belongs to the delta 70 promoter,the prediction success rate is nearly 80%.It is believed that the method proposed in this paper has considerable reference value for the study of promoter prediction.