Identification Of Disease-related Genes And CircRNAs Based On Biological Networks

Complex diseases have high incidence rate and complicated genetic pattern.Besides,they are caused by multiple pathogenic genes or RNA.These factors lead to extremely high mortality rates for complex diseases.Therefore,complex diseases have seriously threatened human health.Prediction of disease-related genes is an urgent problem.Based on the central dogma of genomics,particular regions of DNA are transcribed into RNA.And RNA can be reverse transcribed DNA.Because genes belong to DNA,there is a certain important link between genes and RNA.In other words,RNA is also associated with complex diseases.Therefore,identification of disease-related RNA is also an urgent problem.Identification of disease-related gene and RNA is very pretty believable by traditional biological experiment.However,it will waste large amounts of resources due to vast gene and RNA in human.The computational method to identification disease-related gene and RNA can solve this problem,which because identified disease-related gene and' RNA can be referred to experimenter.In this paper,we propose four computational methods to identify disease-related gene and circRNA based on different biological network.Specific work content is as follows:Firstly,logistic regression algorithm based on reconstructed protein interaction network to identify disease genes.In the first step,in order to solve the problem that traditional PPI networks usually contain a lot of noise,a reconstructed PPI network is constructed based on PPI network and keywords by RWRH algorithm.To analyze gene-disease associations from multiple biological perspectives,a set of heterogeneous features is extracted from the multiple biological data of reconstructed PPI network?protein complex?the gene's tissue expression' keywords and the semantic similarity of genes.And then,a logistic regression algorithm is used to identified disease-related genes.Secondly,identifying Cancer genes by combining two-rounds RWR based on multiple biological data.Biological network,constructing single biological data,ignores the diversity of biological macromolecular functions.Therefore,a big heterogenous network is constructed by PPI network,pathway data?miRNA similarity network?lncRNA similarity network?cancer similarity network and protein complex.Because similar diseases are most likely caused by functionally similar genes,a cancer class is construct by different caner diseases.The first-round of random walk with restart is to identify highly suspicious candidate genes related caner class and delete other genes.And then,a new heterogeneous network is constructed based on identified genes and other biological data.Based on new heterogenous network,the second-round of random walk with restart is employ to identify genes related cancer diseases.Thirdly,predicting circRNA-disease associations based on biased random walk to search paths on a multiple heterogeneous network.Traditional depth-first search algorithm needs to search all paths,which lead to spend a lot of time.Therefore,it is not suitable for large heterogeneous networks.A biased random walk algorithm is proposed to solve this problem.A big heteorgeous network consists of circRNA similarity network?gene similarity network?disease similarity network and their associations.Finally,prediction of circRNA-disease associations based on multiple biological data.Because the research of circRNA is a recent hot topic,there are too few known circRNA-disease associations which cause a negative influence to identify disease-related circRNAs.Therefore,we fuse multiple types of biological data consist of circRNA expression data?disease similarity network and circRNA-disease association to solve this problem.And then,an improved non-negative matrix factorization algorithm is proposed to identify circRNA-disease associations.