An Improved Feature Selection Method And Integrated Analysis Of MiRNA And Gene Networks On Cancer Expression Data

At present,cancer is a major killer to threaten human life and health,so the study of the pathogenesis of cancer has become one of the current research hotspots.If it is possible to determine the incidence of cancer based on the condition of the tumor marker,it is clear that it is beneficial for the prevention and treatment of cancer.Studies have found that differentially expressed genes and miRNAs have the potential as tumor markers to prevent tumors.Mircro RNAs(miRNAs)are one type of small non-coding RNAs and play important roles in regulating wide range of biological processes,such as stem cell maintenance,tissue development and cell metabolism.In the human genome,about 4% of the genes are miRNAs.Abnormal expression of miRNAs usually leads to abnormal expression of the target gene,leading to changes in cell function,such as cell differentiation,proliferation,and even death.Many studies have shown that miRNAs are involved in the biological processes of cancer in humans,so studying the behavior of miRNAs is very important for diagnosing and treating cancer for human life and health.With the development of high-throughput RNASeq technology,aberrant miRNA profiles and gene profiles have been detected in many cancers.Researchers use gene chip technology to obtain a variety of cancer gene and miRNA expression data,and then use these data analysis methods to obtain differentially expressed genes and miRNA,and finally follow the need for subsequent analysis.However,the use of gene chip technology to obtain the data,with high-dimensional small sample characteristics,that is,thousands of genes or miRNA expression data only a few dozen sample data,so that the analysis process to bring some difficulties.In order to solve this problem,generally try to give the data before the dimensionality reduction,feature selection method is the more common data reduction means.The use of excellent feature selection algorithm,that is,the performance of stable and characteristic subset classification of high accuracy algorithm,the follow-up analysis process is very helpful.This paper focuses on the filter feature selection algorithm,which is relatively simple,faster and more stable than other types of feature selection algorithms.In this paper,we improve the maximum correlation minimum redundancy(m RMR)feature selection method.According to recent years,many studies have focused on using pairs of sample data.In order to improve the stability of the algorithm,this paper proposes a more stable feature selection algorithm by using pairs of sample data and integration strategy to improve the algorithm.In the experimental analysis section,six cancer data sets were used to experiment,compared with other feature selection methods,the proposed algorithm has higher accuracy and better stability.In order to study the pathogenesis of cancer,it is important to focus on the abnormal expression of miRNAs and genes and the relationship between them.In this paper,we propose an improved multi-step analysis method to construct a network of miRNAs and targeted genes,and to analyze the relationship between miRNA-targeted gene pairs.First,differentially expressed miRNAs and genes are obtained using improved feature selection algorithms.Then,based on the expression level of the selected miRNAs and genes in the tumor and its corresponding normal tissues,the correlation between the miRNA and the gene was calculated using the Pearson coefficient to obtain a miRNA-targeted gene pair with a negative correlation.Finally,in order to validate our approach,we use three target gene prediction databases to filter the results and analyze them.In the analysis section,the use of four types of cancer chip data,the results of each type of cancer calculated by the database before screening and screening.Through the display of the results and the biological analysis,we can see that we can get some miRNA-targeted pairs that have not appeared in the target gene prediction database using our method.These miRNA-targeted gene pairs can be provided to biological researchers,narrowing the scope for their research work and improving research efficiency.