Predicting lncRNA–miRNA interactions based on logistic matrix factorization with neighborhood regularized

doi:10.1016/j.knosys.2019.105261

Knowledge-Based Systems

Volume 191, 5 March 2020, 105261

https://doi.org/10.1016/j.knosys.2019.105261 Get rights and content

Abstract

Long non-coding RNAs (lncRNAs) and microRNAs (miRNAs) interactions play important roles in diagnostic biomarkers and therapeutic targets for various human diseases. However, experimental methods for finding miRNAs associated with a particular lncRNA are costly, time consuming, and only a few theoretical approaches play a role in predicting potential lncRNA–miRNA associations. In this study, we have established a novel matrix factorization model to predict lncRNA–miRNA interactions, namely lncRNA–miRNA interactions prediction by logistic matrix factorization with neighborhood regularized (LMFNRLMI). Meanwhile, it only utilizes known positive samples to mine potential associations in data that lack negative samples. As a result, this new model obtains reliable performance in the leave-one-out cross validation (the AUC of 0.9319) and 5-fold cross validation (the AUC of 0.9220), which has significantly improved performance in predicting potential lncRNA–miRNA associations compared to other models. Furthermore, comparison with several other network algorithms, and test based on all kinds of similarity, our model successfully confirms the superiority of LMFNRLMI. Whereby, we hope that LMFNRLMI can be a useful tool for potential lncRNA–miRNA association identification in the future.

Introduction

With the rapid development of sequencing technology, a variety of non-coding RNA (ncRNA) data sets [1], [2], [3] have been gradually collected to solve genomic and biological problems. Meanwhile, long non-coding RNAs (lncRNAs) and microRNAs (miRNAs) have received widespread attention during recent years [4], [5]. LncRNAs are a class of ncRNAs with length of more than 200 nucleotides, which have been confirmed to play key roles in complex cellular processes such as epigenetic regulation of gene expression [6], chromatin modification [7] and cell differentiation [8]. In contrast, miRNAs are short regulatory RNAs of approximately 20 nucleotides in length with no protein-coding ability. MiRNAs are also widely involved in a variety of physiological and pathological processes such as development [9], cell proliferation [10], apoptosis and tumorigenesis [11].

A number of studies have reported the effects of miRNAs on lncRNA function [12], [13]. For example, lncRNA–miRNA regulatory networks in gastric cancer [5], colon cancer [9], prostate cancer [14], and vascular disease [15] have been studied. A model for the interaction between lncRNAs and miRNAs has been established and many algorithms can be applied.

The development of new biomarker discovery and therapeutic approaches has been facilitated by a comprehensive and in-depth understanding of the relationship between lncRNA–miRNA and its role in pathophysiology. However, the interaction between lncRNA–miRNAs is still too limited by biological experiments to find the association of lncRNA–miRNAs in large amounts of data. In fact, a quantity of computational methods have been successfully used in assisted biological experiments and are widely used in bioinformatics, such as disease-related relationship prediction [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], drug-target interaction prediction [26] and lncRNA–protein associations prediction [27], [28], [29], [30], [31], [32]. Still, only a few models can be used to forecast lncRNA–miRNA associations. Until recently, Huang et al. [33] first proposed a graph-based method for large-scale prediction of lncRNA–miRNA interactions, which is based on the expression profile similarity to predict potential lncRNA and miRNA associations, ignoring the effects of sequences as essential features of RNA. Using experimental validated data based on sequence similarity, we propose a network-based lncRNA–miRNA interaction prediction model called logistic matrix factorization with neighborhood regularized (LMFNRLMI). Computational models can make discovery more efficient than the complex processes and high costs of experimentation. This algorithm is based on the known interaction information and the similarity of the neighbors to represent the latent factor vector of the matrix factorization into a probability score through the logic function. The idea of k-nearest neighbors is incorporated that k-strongest similarities of the samples are considered and integrated into the regularization items to improve forecasting accuracy. The 5-fold cross validation and leave-one-out cross validation (LOOCV) are used to evaluate our model, with AUC values of 0.9220 and 0.9319, respectively. We evaluate the performance of LMFNRLMI by comparing with other six methods to prove that our method works best. In addition, we use the expression profile similarity and functional similarity of lncRNA and miRNA to verify the generalization ability of the model. The AUC values based on the expression profile similarity and functional similarity are 0.8591 and 0.9131, respectively. The result shows that LMFNRLMI is feasible and effective for lncRNA–miRNA prediction.

Section snippets

Logical matrix factorization

Matrix factorization algorithms have been developed in recommendation systems and have recently applied research on RNA and protein relationship prediction [31]. Here, the observation matrix is decomposed into two low-dimensional matrices $U \times r$ and $V \times r$ to find the inner product, where $r$ is the number of hidden factors. The probabilistic method is used, that is, the probability of occurrence of this event is allocated according to the inner product of the latent factor vector between lncRNA and

Regularized by neighborhood

Although the LMF algorithm can exploit latent vectors to predict potential relationships globally, strong associations of lncRNA–lncRNA and miRNA–miRNA are not considered. Therefore, we propose to use the adjacency matrix of lncRNAs and miRNAs to further improve the prediction accuracy of predicting lncRNA–miRNA interactions by K-nearest neighbors. For example, for lncRNA $l_{i}$ , we use $N (l_{i})$ to represent the nearest neighbor set to lncRNA $l_{i}$ , which contains $k_{1}$ lncRNAs. In the same way, we

Data set

For our investigation, the known lncRNA–miRNA associations are downloaded from lncRNASNP2 database (http://bioinfo.life.hust.edu.cn/lncRNASNP#!/) [37]. lncRNASNP2 integrates alldatabases which is powerful and can provides comprehensive information on SNPs and mutations in lncRNAs, including experimentally validated lncRNA–miRNA pairs, with the latest records reaching 10,597 interactions. 780 different types of lncRNA and 275 different types of miRNAs are included in the records of these

Conclusion

Since the mutual regulation of lncRNA and miRNA in biological mechanism deeply affects human diseases, and experimental verification has high cost and time-consuming limitations, it is an important direction that establishing a lncRNA–miRNA association prediction model in current research. So far, information about lncRNA and miRNA interactions is still limited, but the number of experimental validations is gradually increasing. Relying on reliable experimental data, we build a model called

Acknowledgments

This work was supported by the Doctor Startup Foundation from Liaoning province, China under Grant No. 20170520217, Important Scientific and Technical Achievements Transformation Project, China under Grant No. Z17-5-078, Large-scale Equipment Shared Services Project, China under Grant No. F15165400 and Applied Basic Research Project, China under Grant No. F16205151.

References (41)

YoonJ.H. et al.
Functional interactions among microRNAs and long noncoding RNAs
Semin. Cell Dev. Biol.
(2014)
ZhaoQ. et al.
IRWNRLPI: Integrating random walk and neighborhood regularized logistic matrix factorization for lncrna-protein interaction prediction
Front. Genet.
(2018)
ZhaoQ. et al.
The bipartite network projection-recommended algorithm for predicting long non-coding RNA-protein interactions
Mol. Ther. Nucleic Acids
(2018)
HuH. et al.
HLPI-ensemble: Prediction of human lncRNA-protein interactions based on ensemble strategy
Rna Biol.
(2018)
GengC. et al.
LncRNADisease: a database for long-non-coding RNA-associated diseases
Nucleic Acids Res.
(2013)
VoldersP.J. et al.
LNCipedia: a database for annotated human lncRNA transcript sequences and structures
Nucleic Acids Res.
(2013)
GongJ. et al.
LncRNASNP: a database of SNPs in lncrnas and their potential functions in human and mouse
Nucleic Acids Res.
(2015)
ZhouX. et al.
The interaction between MiR-141 and lncRNA-H19 in regulating cell proliferation and migration in gastric Cancer
Cell. Physiol. Biochem.
(2015)
YeS. et al.
Bioinformatics method to predict two regulation mechanism: TF-miRNA-mRNA and lncRNA-miRNA-mRNA in pancreatic cancer
Cell Biochem. Biophys.
(2014)
OtteP.D.A.P.
Epigenetic regulation of gene expression
J. Braz. Chem. Soc.
(2012)

ResC.

Correction: Long noncoding RNA HOTAIR regulates polycomb-dependent chromatin modification and is associated with poor prognosis in colorectal Cancers

Cancer Res.

(2011)

WangP. et al.

The STAT3-binding long noncoding RNA lnc-DC controls human dendritic cell differentiation

Science

(2014)

BallantyneM.D. et al.

LncRNA/microRNA interactions in the vasculature

Clin. Pharmacol. Ther.

(2016)

LiuB. et al.

Identification of real microrna precursors with a pseudo structure status composition approach

Plos One

(2015)

LiuB. et al.

IMiRNA-PseDPC: microRNA precursor identification with a pseudo distance-pair composition approach

J. Biomol. Struct. Dyn.

(2016)

WangW. et al.

Identification of miRNA lncRNA and mRNA-associated ceRNA networks and potential biomarker for MELAS with mitochondrial DNA A3243G mutation

Sci. Rep.

(2017)

HeJ.H. et al.

Analyzing the lncrna, mirna, and mrna regulatory network in prostate Cancer with bioinformatics software

J. Comput. Biol. J. Comput. Mol. Cell Biol.

(2017)

ZhangC.

MicroRNA-145 in vascular smooth muscle cell biology: A new therapeutic target for vascular disease

Cell Cycle

(2009)

ChenX. et al.

EGBMMDA: Extreme gradient boosting machine for MiRNA-disease association prediction

Cell Death Dis.

(2018)

ChenX. et al.

MicroRNAs and complex diseases: from experimental results to computational models

(2019)

Cited by (105)

Multicomponent comprehensive confirms that erythroferrone is a molecular biomarker of pan-cancer
2024, Heliyon
All vertebrates organisms produce erythroferrone, a secretory hormone with structure-related functions during iron homeostasis. However, limited knowledge exists regarding the effect of this hormone on the occurrence and progression of cancer. To systematically and comprehensively identify the diverse implications of Erythroferrone (ERFE) in various malignant tumors, we conducted an in-depth analysis of multiple datasets, including the expression levels of oncogenes and target proteins, biological functions, and molecular characteristics. This analysis aimed to assess the diagnostic and prognostic value of ERFE in pan-cancer. Our findings revealed a significant elevation in ERFE expression across 20 distinct cancer types, with notable increases in gastrointestinal cancers. Utilizing the Cytoscape and STRING databases, we identified 35 ERFE-targeted binding proteins. Survival prognosis studies, particularly gastrointestinal cancers indicated by Colon adenocarcinoma (COAD), demonstrated a poor prognosis in patients with high ERFE expression (p < 0.001), consistently observed across various clinical subgroups. Furthermore, the ROC curve underscored the high predictive ability of EFRE for gastrointestinal cancer (AUC >0.9). Understanding the roles and interactions of ERFE in biological processes can also be aided by examining the genes co-expressed with ERFE in the coat and ranking the top 50 positive and negative genes. In the correlation analysis between the ERFE gene and different immune cells in COAD, we discovered that the expression of ERFE was positively correlated with Th1 cells, cytotoxic cells, and activated DC (aDC) abundance, and negatively correlated with Tcm (T central memory) abundance (P < 0.001). in summary, ERFE emerges as strongly associated with various malignant cancers, positioning it as a prospective biological target for cancer treatment. It stands out as a key molecular biomarker for diagnosing and prognosticating pancreatic cancer, also serves as an independent prognostic risk factor for COAD.
scAAGA: Single cell data analysis framework using asymmetric autoencoder with gene attention
2023, Computers in Biology and Medicine
In recent years, single-cell RNA sequencing (scRNA-seq) has emerged as a powerful technique for investigating cellular heterogeneity and structure. However, analyzing scRNA-seq data remains challenging, especially in the context of COVID-19 research. Single-cell clustering is a key step in analyzing scRNA-seq data, and deep learning methods have shown great potential in this area. In this work, we propose a novel scRNA-seq analysis framework called scAAGA. Specifically, we utilize an asymmetric autoencoder with a gene attention module to learn important gene features adaptively from scRNA-seq data, with the aim of improving the clustering effect. We apply scAAGA to COVID-19 peripheral blood mononuclear cell (PBMC) scRNA-seq data and compare its performance with state-of-the-art methods. Our results consistently demonstrate that scAAGA outperforms existing methods in terms of adjusted rand index (ARI), normalized mutual information (NMI), and adjusted mutual information (AMI) scores, achieving improvements ranging from 2.8% to 27.8% in NMI scores. Additionally, we discuss a data augmentation technology to expand the datasets and improve the accuracy of scAAGA. Overall, scAAGA presents a robust tool for scRNA-seq data analysis, enhancing the accuracy and reliability of clustering results in COVID-19 research.
MPFFPSDC: A multi-pooling feature fusion model for predicting synergistic drug combinations
2023, Methods
Drug combination therapies are common practice in the treatment of cancer, but not all combinations result in synergy. As traditional screening approaches are restricted in their ability to uncover synergistic drug combinations, computer-aided medicine is becoming a increasingly prevalent in this field. In this work, a predictive model of potential interactions between drugs named MPFFPSDC is presented, which can maintain the symmetry of drug inputs and eliminate inconsistencies in predictive results caused by different drug inputting sequences or positions. The experimental results show that MPFFPSDC outperforms comparative models in major performance indicators and exhibits better generalization for independent data. Furthermore, the case study demonstrates that our model can capture molecular substructures that contribute to the synergistic effect of two drugs. These results indicate that MPFFPSDC not only offers strong predictive performance, but also has good model interpretability that may provide new insights for the study of drug interaction mechanisms and the development of new drugs.
Multi-view graph neural network with cascaded attention for lncRNA-miRNA interaction prediction
2023, Knowledge-Based Systems
Identifying interactions between long non-coding RNAs (lncRNAs) and microRNAs (miRNAs) reveals the mechanisms of biological processes, thereby contributing to disease diagnosis and treatment. Recently, graph neural networks (GNNs) have achieved remarkable progress in this task due to their consideration of both node attributes and graph topology. Nevertheless, existing GNN-based methods use only one type of node attribute, and the possible bias of a single view leads them to learn suboptimal node representations. Moreover, the underlying mechanisms of action between lncRNAs and miRNAs are complex. Ignoring the importance of neighboring nodes to the target node and the influence of different order neighborhood information makes them fail to learn satisfactory topological information. To this end, we propose a novel Multi-view Graph Neural Network with Cascaded ATtention (MGCAT) for lncRNA-miRNA interaction (LMI) prediction, where cascaded attention is a key ingredient consisting of view-level, node-level, and layer-level attentions. Specifically, we first construct a multi-attributed LMI graph to fully characterize lncRNAs and miRNAs, where nodes have multiple node attributes (i.e., multi-view features). Next, view-level attention dynamically integrates multi-view features to capture the inherent attribute information of nodes. Then, node-level attention iteratively aggregates the neighborhood information of each node. Finally, layer-level attention adaptively combines integrated features and different order neighborhood information to obtain informative node representations. Extensive experiments on four benchmark datasets show that MGCAT consistently outperforms recent state-of-the-art methods. Further case studies demonstrate the potential ability of MGCAT to identify novel LMIs. Code and datasets are publicly available at https://github.com/ai4slab/mgcat.
Self-representative kernel concept factorization
2023, Knowledge-Based Systems
Kernel concept factorization (KCF) has successfully utilized kernel trick to conduct matrix factorization in the kernel space. However, conventional KCF methods usually define kernel in advance, which limits their ability to exploit the power of kernel. This paper proposes a semi-supervised self-representative kernel concept factorization (S $^{3}$ RKCF) method to integrate adaptive kernel learning and low-dimensional data representation learning into a unified model. Technically, an adaptive local geometric structure is acquired in the KCF-induced self-representation space, and then it facilitates data representation learning simultaneously. Furthermore, to enhance the discriminability of data representation, limited supervisory information is imposed in the formulated optimization problem as constraints. In this way, our model can learn kernel and discriminative low-dimensional representation adaptively and iteratively. To solve the optimization problem, an alternating iterative algorithm is designed with convergence guarantee. The performance of our proposed S $^{3}$ RKCF is evaluated through clustering and classification tasks, and the experimental results on eight real-world data sets demonstrate its effectiveness compared to state-of-the-art methods.
Predicting miRNA-disease association through combining miRNA function and network topological similarities based on MINE
2022, iScience
Predicting associations between microRNAs (miRNAs) and diseases from the viewpoint of function modules has become increasingly popular. However, existing methods obtained the relations between diseases and miRNAs only through the construction of similarity networks and neglected the complex network characteristic. In this paper, a new method named combining miRNA function similarities and network topology similarities based on module identification in networks (ComSim-MINE) was developed. Combined similarity is calculated from the harmonic mean between miRNA function similarities and network topology similarities. Experimental results showed that ComSim-MINE can compete with several state-of-the-art weighted function module algorithms, such as ClusterONE, MCODE, NEMO, and SPICi, and achieved the satisfactory results in terms of the composite score of F-measure, sensitivity, and accuracy based on the generated miRNA function interaction network. From the analysis of case studies, some new findings obtained from our proposed method provide clinicians new clues for epidemic diseases, such as COVID-19.

View all citing articles on Scopus

^☆: No author associated with this paper has disclosed any potential or pertinent conflicts which may be perceived to have impending conflict with this work. For full disclosure statements refer to https://doi.org/10.1016/j.knosys.2019.105261.

View full text

Predicting lncRNA–miRNA interactions based on logistic matrix factorization with neighborhood regularized☆

Abstract

Introduction

Section snippets

Logical matrix factorization

Regularized by neighborhood

Data set

Conclusion

Acknowledgments

Semin. Cell Dev. Biol.

Front. Genet.

Mol. Ther. Nucleic Acids

Rna Biol.

LncRNADisease: a database for long-non-coding RNA-associated diseases

Nucleic Acids Res.

LNCipedia: a database for annotated human lncRNA transcript sequences and structures

Nucleic Acids Res.

LncRNASNP: a database of SNPs in lncrnas and their potential functions in human and mouse

Nucleic Acids Res.

The interaction between MiR-141 and lncRNA-H19 in regulating cell proliferation and migration in gastric Cancer

Cell. Physiol. Biochem.

Bioinformatics method to predict two regulation mechanism: TF-miRNA-mRNA and lncRNA-miRNA-mRNA in pancreatic cancer

Cell Biochem. Biophys.

Epigenetic regulation of gene expression

J. Braz. Chem. Soc.

Correction: Long noncoding RNA HOTAIR regulates polycomb-dependent chromatin modification and is associated with poor prognosis in colorectal Cancers

Cancer Res.

The STAT3-binding long noncoding RNA lnc-DC controls human dendritic cell differentiation

Science

LncRNA/microRNA interactions in the vasculature

Clin. Pharmacol. Ther.

Identification of real microrna precursors with a pseudo structure status composition approach

Plos One

IMiRNA-PseDPC: microRNA precursor identification with a pseudo distance-pair composition approach

J. Biomol. Struct. Dyn.

Identification of miRNA lncRNA and mRNA-associated ceRNA networks and potential biomarker for MELAS with mitochondrial DNA A3243G mutation

Sci. Rep.

Analyzing the lncrna, mirna, and mrna regulatory network in prostate Cancer with bioinformatics software

J. Comput. Biol. J. Comput. Mol. Cell Biol.

MicroRNA-145 in vascular smooth muscle cell biology: A new therapeutic target for vascular disease

Cell Cycle

EGBMMDA: Extreme gradient boosting machine for MiRNA-disease association prediction

Cell Death Dis.

MicroRNAs and complex diseases: from experimental results to computational models