LOMDA: Linear optimization for miRNA-disease association prediction

MicroRNAs (miRNAs) have been playing a crucial role in many important biological processes. Currently, the validated associations between miRNAs and diseases are insufficient comparing to all underlying associations. To identify these hidden associations based on biological experiment is expensive, laborious and time consuming. Therefore, computationally inferring the potential associations from biological data for further biological experiment has attracted increasing interests from different communities ranging from biological to computational science. In this work, we propose an effective and flexible method to predict the associations between miRNAs and diseases, namely linear optimization (LOMDA). The proposed method is capable of predicting the associations in three manners e.g., extra information such as miRNA functional similarity, gene functional similarity and known miRNA-disease associations are available; only some associations are known; and new miRNAs or diseases that do not have any known associations at all. The average AUC obtained from LOMDA over 15 diseases in a 5-fold-cross validation is 0.997, while the AUC of 5-fold cross validation on all diseases is 0.957. Moreover, the average AUC on leave-one-out cross validation is 0.866. We compare LOMDA with the state-of-the-art methods and the results show that LOMDA outperforms the others in both cases, e.g., extra information is combined and only known associations are used.