LRMDA: Using Logistic Regression and Random Walk with Restart for MiRNA-Disease Association Prediction

MicroRNAs (MiRNAs) have received much attention in recent years because growing evidences indicate that they play critical roles in tumor initiation and progression. Predicting underlying disease-related miRNAs from existing huge amount of biological data is a hot topic in biomedical research. Herein, we presented a novel computational model of logistic regression and random walk with restart algorithm for miRNA-disease association prediction (LRMDA) through integrating multi-source data. The model employs random walk with restart to fuse the association distribution between miRNAs and diseases and obtains highly discriminative feature from those heterogeneous data. To evaluate the performance of LRMDA, we performed 5-fold cross validation to compare it with several state-of-the-art models. As a result, our model achieves mean AUC of 0.9230 ± 0.0059. Besides, we carried out case study for predicting potential miRNAs related to Esophageal Neoplasms (EN). The achieved results indicate that 90% out of the top 50 prioritized miRNAs for EN are confirmed by biological experiments and further demonstrates the feasibility of our method. Therefore, LRMDA could potentially aid future research efforts for miRNA-disease association identification.