Detection Model And Algorithms For Pathogenic Snps And Genes

Genome-wide association study mainly focuses on scanning and sequencing of genome,to find a phenotype or a disease of single nucleotide polymorphisms and mutations in the context of the whole genome.In recent years,a number of algorithms have been proposed to extract the correlations in the field.Although the algorithms have achieved some success,these studies have pointed out that they show a certain ambiguity in general data.Therefore,in view of the above research purposes,the following research works are carried out:To detect potentially pathogenic single nucleotide polymorphism,our methods for single nucleotide polymorphism detection are proposed.The first one is the chi square test model based on the Bonferroni correction.Compared with the traditional Chi square test,the model greatly reduces the false positive results of the test results.Another is the maximum MICSNP(Maximal Information Coefficien Single Nucleotide Polymorphisms)that based on the maximum information coefficient,MICSNP algorithm has good universality in the premise of unknown function model,and is able to detect disease associated polymorphisms.In addition,the two methods are compared.It is found that the cross validation of the results of the two methods can further improve the accuracy of the results.Potentially pathogenic gene testing.Genes can be considered as a collection of loci that are often the result of interactions between internal sites.In this paper,deep learning LPGDM model is proposed based on the detection of pathogenic gene(LSTM Pathogenic Gene Detection Model),LPGDM model with long term memory network as the core,to overcome the problem of long-term dependence on traditional recurrent neural network,and discover the relationship between the site and the gene level of abstraction.In order to verify the detection effect of the LPGDM model,the LPGDM model and the traditional machine learning model(SVM,Decision Tree,Logistic Regression,Naive Bayesian)were compared,and we found that the LPGDM model based on deep learning in the pathogenic gene detection effect is superior to the traditional machine learning model.