Prodiction Of Cancer-related LncRNAs Based On Fusion Characteristics

Increasing evidences reveal that long noncoding RNAs(lncRNAs)play an important role in many human diseases.However,it is time-consuming and labor-intensive to find disease-associated long noncoding RNAs based on traditional experimental methods.In recent years,the effective identification of the potential functions of long noncoding RNAs and the recognition of RNAs associated with diseases have studied by using computational models.In this paper,based on k-mer information and GC frequency of lncRNAs sequence,number of stem loops of lncRNAs secondary structure,geometric flexible information of sequence,the number of repeat elements in the lncRNAs,and combined with the relationship between cancer-related mi RNAs and lncRNAs,the cancer-associated lncRNAs were predicted by using support vector machine algorithm.It was found that best predictive results are obtained by using fusing k-mer information,flexible information of lncRNAs sequence and the relationship of cancer-related miRNAs and lncRNAs.The total prediction accuracies reached 82.36% and 77.59% in Jackknife test and independent test,respectively.The results indicate that k-mer information of the sequence,geometric flexible information of lncRNAs and the interaction between cancer-related miRNAs and lncRNAs are very useful for predicting cancer-associated lncRNAs.