Search

MicroRNAs (miRNAs) play a key role in many critical biological processes and are involved in the occurrence and development of complex human diseases. Many studies demonstrated that discovering the associations between small molecules (SMs) and miRNAs will facilitate the design of miRNA targeted therapeutic strategies for complex human diseases. This work presents a calculation model of cross-layer dependency inference on multilayered networks for small molecule-miRNA association prediction (CLDISMMA), which constructed multilayered networks composed of SMs, miRNAs, and diseases. It utilized the within layer topology and the known cross-layer associations to infer latent representations of all layers for SM-miRNA association prediction. In CLDISMMA, the novelties lie in introducing disease information for SM-miRNA association prediction and utilizing a regularized optimization model to describe the SM-miRNA association prediction problem. To evaluate the performance of CLDISMMA, global leave-one-out cross validation (LOOCV) and miRNA-fixed and SM-fixed local LOOCV were implemented in two data sets. In data set 1, CLDISMMA achieved AUCs of 0.9889, 0.9886, and 0.7755 in turns. The corresponding AUCs were 0.8726, 0.8798, and 0.7021 based on data set 2. In addition, CLDISMMA obtained average AUCs of 0.9887 and 0.8647 in data sets 1 and 2 under 100 times 5-fold cross validation. Furthermore, we employed CLDISMMA to predict SM-miRNA associations based on data set 1, and 21 out of the top 50 predicted associations were confirmed by experimental reports. In the case study for new SMs, 5-fluorouracil and 5-aza-2'-deoxycytidine, 40 and 30 miRNAs, respectively, were verified to be associated with them among the top 50 miRNAs predicted by CLDISMMA.

More and more studies found that many complex human diseases occur accompanied by aberrant expression of microRNAs (miRNAs). Small molecule (SM) drugs have been utilized to treat complex human diseases by affecting the expression of miRNAs. Several computational methods were proposed to infer underlying associations between SMs and miRNAs. In our study, we proposed a new calculation model of random forest based small molecule-miRNA association prediction (RFSMMA) which was based on the known SM-miRNA associations in the SM2miR database. RFSMMA utilized the similarity of SMs and miRNAs as features to represent SM-miRNA pairs and further implemented the machine learning algorithm of random forest to train training samples and obtain a prediction model. In RFSMMA, integrating multiple kinds of similarity can avoid the bias of single similarity and choosing more reliable features from original features can represent SM-miRNA pairs more accurately. We carried out cross validations to assess predictive accuracy of RFSMMA. As a result, RFSMMA acquired AUCs of 0.9854, 0.9839, 0.7052, and 0.9917 ± 0.0008 under global leave-one-out cross validation (LOOCV), miRNA-fixed local LOOCV, SM-fixed local LOOCV, and 5-fold cross validation, respectively, under data set 1. Based on data set 2, RFSMMA obtained AUCs of 0.8456, 0.8463, 0.6653, and 0.8389 ± 0.0033 under four cross validations according to the order mentioned above. In addition, we implemented a case study on three common SMs, namely, 5-fluorouracil, 17β-estradiol, and 5-aza-2'-deoxycytidine. Among the top 50 associated miRNAs of these three SMs predicted by RFSMMA, 31, 32, and 28 miRNAs were verified, respectively. Therefore, RFSMMA is shown to be an effective and reliable tool for identifying underlying SM-miRNA associations.