Meta-analysis Of Differentially Expressed Genes And Its Application In Citrus HuangLongBing

Citrus Huanglongbing,known as citrus cancer,is a devastating disease of citrus,which seriously restricts the development of global citrus industry.With the development of molecular biology,people’s understanding of Citrus Huanglongbing opened a new cognitive world.By using the differentially expressed genes of citrus,the biological process and pathogenic mechanism of Citrus Huanglongbing can be more accurately defined,and the corresponding susceptible genes can be excavated,so as to gain insight into the expression process of Citrus Huanglongbing.To solve the above problems,this paper used meta-analysis to study the differentially expressed genes of Citrus Huanglongbing.In view of the large number of differentially expressed genes in citrus,the combination effect value method was considered to determine the susceptibility genes of differentially expressed genes according to the size of the effect value.We designed a meta-analysis method based on combination value and P value to study the differentially expressed genes in Citrus Huanglongbing.Firstly,a meta-analysis model of differentially expressed genes in Citrus Huanglongbing was established based on combination value and P value,and then the data were preprocessed to obtain the effect value of the corresponding genes,and the experiments were compared and analyzed.The experimental results showed that there were 5 negative genes and 44 positive genes.According to the comparison of ROC curves,the AUC of this method was 0.950.In order to solve the problem that the data set of differentially expressed genes in citrus may be overfitted,a meta-analysis model of differentially expressed genes in Citrus Huanglongbing based on random forest was designed and established to study the differentially expressed genes.Firstly,a meta-analysis model of differential expression genes of Citrus Huanglongbing based on random forest was established,and then the data were preprocessed,and the corresponding results were obtained and the experiments were compared.The experimental results showed that there were 27 negative genes and 77 positive genes.According to the comparison of ROC curves,the AUC of this method was 0.980.The overlap rate of the two methods was calculated by comparing the combination effect and P-value meta-analysis method and the random forest-based meta-analysis method.The overlap rate of negative genes based on the combination effect and P-value meta-analysis method was80%,while that based on the random forest-based meta-analysis method was 14.29%.The coincidence rate of positive genes was 68.18% based on meta-analysis of combinatorial effects and 38.96% based on meta-analysis of random forest.The results suggested that the combination effect and p-value meta-analysis method could obtain more result recognition and have higher credibility in the analysis of differential expression genes of Citrus Huanglongbing.