Study On Algorithm To Explore LncRNA Related Regulatory Network Of Breast Cancer

The causes of breast cancer are complex and are closely related to genetic inheritance,growth environment or living habits,among which whether the gene is expressed normally is the most critical factor for breast cancer.Studying the regulatory networks of genes can provide a new research perspective for the mechanism of breast cancer.There are many ways to study gene regulatory networks,including computational methods to identify regulatory networks which is widely used and developed due to its excellent performance.Protein-coding genes in the human genome account for about 1%?2% and the rest are non-coding RNA(ncRNA).LncRNA is composed of more than 200 nucleotides,accounting for 80% to 90% of non-coding RNA.In addition,studies have shown that lncRNA can be used as a kind of competitive endogenous RNA(ce RNA),widely involved in cell cycle regulation,signaling pathways,hormone regulation,and breast cancer development.In this paper,we proposed two new methods to identify lncRNA-related regulatory networks in breast cancer.There are presented as following:(1)A new algorithm called EPMSI based on matrix completion to explore lncRNA-lncRNA and lncRNA-mRNA regulatory networks in breast cancer has been proposed.Based on the limited known regulatory network and gene expression of breast cancer,EPMSI has explored 3 000 another potential lncRNA-mRNA and 10 000 another mRNA-mRNA regulatory networks which are rich in biological enrichments and each pair of identified association pair has an inferred value to signify interaction intensity.When BCPlaid,a clustering algorithm,is applied to regulatory networks.10 regulatory modules are identified,of which 9 regulatory modules are closely related to pathways closely to breast cancer.When SVM with ten-fold cross-validation is used to distinguish normal and tumor samples for each regulatory module,10 modules have shown excellent performance.All facts show that the regulatory network identified by EPMSI is closely related to breast cancer,and may be a promising method for researchers to study the pathogenesis of breast cancer.(2)An algorithm called MSNMDL has been proposed to predict the lncRNAmRNA regulatory network of breast cancer.MSNMDL first explores the miRNAlncRNA functional similarity network from the miRNA-disease-lncRNA network.And then MSNMDL replaces the corresponding miRNA in the miRNA-target network with the lncRNA of each pair of lncRNA-miRNA in the miRNA-lncRNA functional similarity network,thus lncRNA-target regulatory networks are formed.When enrichment analysis,survival analysis,regulatory module analysis,disease enrichment analysis,verification and experimental comparison are performed on indentified regulatory network,the results show that MSNMDL significantly improves the performance including bioinformatics significance and perhaps its identified module may serve as marker of regulatory network for breast cancer prognosis.