ncRNA-disease association prediction based on sequence information and tripartite network

Abstract

Background: Current technology has demonstrated that mutation and deregulation of non-coding RNAs (ncRNAs) are associated with diverse human diseases and important biological processes. Therefore, developing a novel computational method for predicting potential ncRNA-disease associations could benefit pathologists in understanding the correlation between ncRNAs and disease diagnosis, treatment, and prevention. However, only a few studies have investigated these associations in pathogenesis.

Results: This study utilizes a disease-target-ncRNA tripartite network, and computes prediction scores between each disease-ncRNA pair by integrating biological information derived from pairwise similarity based upon sequence expressions with weights obtained from a multi-layer resource allocation technique. Our proposed algorithm was evaluated based on a 5-fold-cross-validation with optimal kernel parameter tuning. In addition, we achieved an average AUC that varies from 0.75 without link cut to 0.57 with link cut methods, which outperforms a previous method using the same evaluation methodology. Furthermore, the algorithm predicted 23 ncRNA-disease associations supported by other independent biological experimental studies.

Conclusions: Taken together, these results demonstrate the capability and accuracy of predicting further biological significant associations between ncRNAs and diseases and highlight the importance of adding biological sequence information to enhance predictions.

Keywords: Resource allocation; Tripartite network; ncRNA-disease association predictions.

Fig. 1

Cumulative degree distribution of the tripartite ncRNA-target disease network for Chen et al. dataset

Fig. 2

Example of a tripartite network used in this work. Targets integrate information from ncRNAs and diseases

Fig. 3

Illustration of the proposed computational method. Sequence information analysis is combined with a tripartite network structure. A multi-layer resource-allocation technique integrates the information and predicts associations between ncRNAs and human diseases

Fig. 4

Algorithm performance evaluation methodology using 5-fold cross validation. From the constructed network as shown in (a), we divide all ncRNA-disease pairs into 5 groups. Each time one group performs as a testing dataset. To evaluate whether the algorithm is able to predict the interacting pairs without memory of existing interactions, we delete links that connect ncRNAs with disease if they belong to the testing set as shown in (b)

Fig. 5

Comparison between the proposed method (MRAS) and ncPred. a shows an ROC curves comparison between the proposed method (MRAS) and ncPred with evaluation of the Chen et al. dataset. The ROC curves were drawn up based on the average results of the simulation, and repeated to ensure reliable estimates as described in the evaluation method procedure. The results also convey that the proposed method produced a higher true positive rate, which demonstrates its superior performance over its competitor. b shows the results of the proposed method using different kernel functions

Fig. 6

Illustration of the ROC curves for the proposed algorithm. ROC curves for the proposed algorithm MRAS using different values of l-gram string (l = 1, 2, 3 and 4) and using RBF kernel (a) without link cut (AUC = 0.75) and (b) with link cut (AUC = 0.57) for the Chen dataset. The results for the Helwak dataset using RBF kernel (c) without link cut (AUC = 0.85) and (d) with link cut (AUC = 0.77)