---
title: "miRsponge: identification and analysis of miRNA sponge 
    interaction networks and modules" 
author: "\\
    
    Junpeng Zhang (<zhangjunpeng_411@yahoo.com>)\\
    
    School of Engineering, Dali University"
date: '`r Sys.Date()`'
output:
    BiocStyle::html_document:
      toc: yes
    BiocStyle::pdf_document:
      toc: yes
vignette: >
    %\VignetteIndexEntry{miRsponge: identification and analysis of miRNA sponge interaction networks and modules} 
    %\VignettePackage{miRsponge} 
    % \VignetteEngine{knitr::rmarkdown} 
    % \usepackage[utf8]{inputenc} 
    % \VignetteEncoding{UTF-8}
---

```{r style, echo=FALSE, results="asis", message=FALSE}
BiocStyle::markdown()
knitr::opts_chunk$set(tidy = FALSE,
    warning = FALSE,
    message = FALSE)
```

```{r echo=FALSE, results='hide', message=FALSE}
library(miRsponge)
```


# Introduction
MicroRNAs (miRNAs) are small non-coding RNAs (~22 nt) that control a 
wide range of biological processes including cancers via regulating 
target genes [1-5]. Therefore, it is important to uncover miRNA functions 
and regulatory mechanisms in cancers. 

The emergence of competing endogenous RNA (ceRNA) hypothesis [6] 
has challenged the traditional knowledge that coding RNAs only act 
as targets of miRNAs. Actually, a pool of coding and non-coding 
RNAs that shares common miRNA biding sites competes with each other, 
thus act as ceRNAs to release coding RNAs from miRNAs control. 
These ceRNAs are also known as miRNA sponges or miRNA decoys, and 
include long non-coding RNAs (lncRNAs), pseudogenes, circular RNAs 
(circRNAs) and messenger RNAs (mRNAs), etc [7-10]. Recent 
studies [11, 12] have shown that miRNA sponge network and module 
can help to reveal the biological mechanism in cancer.

To accelerate the research of miRNA sponge, we develop an R package 
'miRsponge' to implement popular methods in the identification and 
analysis of miRNA sponge network and module.
    
# Identification of miRNA sponge interactions
In 'spongeMethod' function, We implement seven popular methods 
(miRHomology [13, 14], pc [15, 16], sppc [17], ppc [11], hermes [18], 
muTaME [19], and cernia [20]) to identify miRNA sponge interactions. 
The seven methods should meet a basic condition: the significance of
sharing of miRNAs by each RNA-RNA pair (e.g. adjusted p-value < 0.01).
Each method has its own merit due to different evaluating indicators. 
Thus, we present an integrate method to combine predicted miRNA
sponge interactions from different methods. 

## miRHomology

We implement miRHomology method based on the 
homology of miRNA binding sites present in the 3'UTR sequence.
```{r}
miR2Target <- system.file("extdata", "miR2Target.csv", package="miRsponge")
miRTarget <- read.csv(miR2Target, header=TRUE, sep=",")
miRHomologyceRInt <- spongeMethod(miRTarget, method = "miRHomology")
head(miRHomologyceRInt)
```


## pc

The pc function considers expression data based on miRHomology method. 
Significantly positive miRNA sponge interaction pairs are regarded as output. 

```{r}
miR2Target <- system.file("extdata", "miR2Target.csv", package="miRsponge")
miRTarget <- read.csv(miR2Target, header=TRUE, sep=",")
ExpDatacsv <- system.file("extdata", "ExpData.csv", package="miRsponge")
ExpData <- read.csv(ExpDatacsv, header=FALSE, sep=",")
pcceRInt <- spongeMethod(miRTarget, ExpData, method = "pc")
head(pcceRInt)
```


## sppc

We implement sppc method based on sensitivity 
correlation (difference between the Pearson and partial 
correlation coefficients).

```{r}
miR2Target <- system.file("extdata", "miR2Target.csv", package="miRsponge")
miRTarget <- read.csv(miR2Target, header=TRUE, sep=",")
ExpDatacsv <- system.file("extdata", "ExpData.csv", package="miRsponge")
ExpData <- read.csv(ExpDatacsv, header=FALSE, sep=",")
sppcceRInt <- spongeMethod(miRTarget, ExpData, senscorcutoff = 0.1, method = "sppc")
head(sppcceRInt)
```


## hermes

The hermes method predicts competing endogenous RNAs 
via evidence for competition for miRNA regulation based on conditional 
mutual information.

```{r}
miR2Target <- system.file("extdata", "miR2Target.csv", package="miRsponge")
miRTarget <- read.csv(miR2Target, header=TRUE, sep=",")
ExpDatacsv <- system.file("extdata", "ExpData.csv", package="miRsponge")
ExpData <- read.csv(ExpDatacsv, header=FALSE, sep=",")
hermesceRInt <- spongeMethod(miRTarget, ExpData, num_perm = 10, method = "hermes")
head(hermesceRInt)
```

Parameter 'num_perm' is used to set the number of permutations of 
input expression data. The larger the number is, the slower the calculation is.


## ppc

The ppc method is a variant of the hermes method. 
However, it predicts competing endogenous RNAs via evidence for competition 
for miRNA regulation based on Partial Pearson Correlation.

```{r}
miR2Target <- system.file("extdata", "miR2Target.csv", package="miRsponge")
miRTarget <- read.csv(miR2Target, header=TRUE, sep=",")
ExpDatacsv <- system.file("extdata", "ExpData.csv", package="miRsponge")
ExpData <- read.csv(ExpDatacsv, header=FALSE, sep=",")
ppcceRInt <- spongeMethod(miRTarget, ExpData, num_perm = 10, method = "ppc")
head(ppcceRInt)
```

Parameter 'num_perm' is used to set the number of permutations of input 
expression data. The larger the number is, the slower the calculation is.

## muTaME

We implement the muTaME method based on the logarithm 
of four scores: (1) the fraction of common miRNAs, (2) the density of the 
MREs for all shared miRNAs, (3) the distribution of MREs of the putative 
RNA-RNA pairs and (4) the relation between the overall number of MREs for 
a putative miRNA sponge compared with the number of miRNAs that yield these 
MREs. There is no reason to decide which score has more contribution than 
the rest. Thus, we calculate a combined score by adding these four scores. 
To evaluate the strength of each RNA-RNA pair, we further normalize the combined 
scores to obtain normalized scores with interval [0 1].

```{r}
miR2Target <- system.file("extdata", "miR2Target.csv", package="miRsponge")
miRTarget <- read.csv(miR2Target, header=TRUE, sep=",")
MREs <- system.file("extdata", "MREs.csv", package="miRsponge")
mres <- read.csv(MREs, header=TRUE, sep=",")
muTaMEceRInt <- spongeMethod(miRTarget, mres = mres, method = "muTaME")
head(muTaMEceRInt)
```


## cernia

We implement the cernia method based on the 
logarithm of seven scores: (1) the fraction of common miRNAs, (2) 
the density of the MREs for all shared miRNAs, (3) the distribution 
of MREs of the putative RNA-RNA pairs, (4) the relation between the 
overall number of MREs for a putative miRNA sponge compared with the 
number of miRNAs that yield these MREs, (5) the density of the 
hybridization energies related to MREs for all shared miRNAs, (6) 
the DT-Hybrid recommendation scores and (7) the pairwise Peason 
correlation between putative RNA-RNA pair expression data. There 
is no reason to decide which score has more contribution than the 
rest. Thus, we calculate a combined score by adding these seven scores. 
To evaluate the strength of each RNA-RNA pair, we further normalize 
the combined scores to obtain normalized scores with interval [0 1].

```{r}
miR2Target <- system.file("extdata", "miR2Target.csv", package="miRsponge")
miRTarget <- read.csv(miR2Target, header=TRUE, sep=",")
MREs <- system.file("extdata", "MREs.csv", package="miRsponge")
mres <- read.csv(MREs, header=TRUE, sep=",")
ExpDatacsv <- system.file("extdata", "ExpData.csv", package="miRsponge")
ExpData <- read.csv(ExpDatacsv, header=FALSE, sep=",")
cerniaceRInt <- spongeMethod(miRTarget, ExpData, mres, method = "cernia")
head(cerniaceRInt)
```


## integrateMethod

To obtain a list of high-confidence miRNA sponge interactions, we 
implement 'integrateMethod' function to integrate results of different methods. 

```{r}
miR2Target <- system.file("extdata", "miR2Target.csv", package="miRsponge")
miRTarget <- read.csv(miR2Target, header=TRUE, sep=",")
ExpDatacsv <- system.file("extdata", "ExpData.csv", package="miRsponge")
ExpData <- read.csv(ExpDatacsv, header=FALSE, sep=",")
miRHomologyceRInt <- spongeMethod(miRTarget, method = "miRHomology")
pcceRInt <- spongeMethod(miRTarget, ExpData, method = "pc")
sppcceRInt <- spongeMethod(miRTarget, ExpData, method = "sppc")
Interlist <- list(miRHomologyceRInt[, 1:2], pcceRInt[, 1:2], sppcceRInt[, 1:2])
IntegrateceRInt <- integrateMethod(Interlist, Intersect_num = 2)
head(IntegrateceRInt)
```

Parameter 'Intersect_num' is used to set the least number of methods 
intersected for integration. That is to say, we only reserve those 
miRNA sponge interactions predicted by at least 'Intersect_num' methods. 

# Validation of miRNA sponge interactions

To validate the predicted miRNA sponge interactions, we implement 
'spongeValidate' function. The groundtruth of miRNA sponge interactions 
are from miRSponge (<http://www.bio-bigdata.net/miRSponge/>) and the experimentally validated miRNA sponge interactions of related literatures.

```{r}
miR2Target <- system.file("extdata", "miR2Target.csv", package="miRsponge")
miRTarget <- read.csv(miR2Target, header=TRUE, sep=",")
miRHomologyceRInt <- spongeMethod(miRTarget, method = "miRHomology")
Groundtruthcsv <- system.file("extdata", "Groundtruth.csv", package="miRsponge")
Groundtruth <- read.csv(Groundtruthcsv, header=TRUE, sep=",")
spongenetwork_validated <- spongeValidate(miRHomologyceRInt[, 1:2], directed = FALSE, Groundtruth)
spongenetwork_validated
```


# Module identification from miRNA sponge interaction network

To further understand the module-level properties of miRNA sponges in 
cancer, we implement 'netModule' function to identify miRNA sponge modules. 
Users can choose FN [21], MCL [22], LINKCOMM [23] and MCODE [24] for module identification.

```{r}
miR2Target <- system.file("extdata", "miR2Target.csv", package="miRsponge")
miRTarget <- read.csv(miR2Target, header=TRUE, sep=",")
miRHomologyceRInt <- spongeMethod(miRTarget, method = "miRHomology")
spongenetwork_Cluster <- netModule(miRHomologyceRInt[, 1:2], modulesize = 2)
spongenetwork_Cluster
```

# Disease and functional enrichment analysis of miRNA sponge modules
We implement 'moduleDEA' function to make disease enrichment analysis of 
modules. The disease databases used include DO: Disease 
Ontology database (<http://disease-ontology.org/>), 
DGN: DisGeNET database (<http://www.disgenet.org/>) 
and NCG: Network of Cancer Genes database 
(<http://ncg.kcl.ac.uk/>). Moreover, 'moduleFEA' function 
is implemented to conduct functional GO, KEGG and Reactome 
enrichment analysis of modules. The ontology databases used
contain GO: Gene Ontology database 
(<http://www.geneontology.org/>), KEGG: 
Kyoto Encyclopedia of Genes and Genomes Pathway Database 
(<http://www.genome.jp/kegg/>), and Reactome:
Reactome Pathway Database (<http://reactome.org/>).

```{r, eval=FALSE, include=TRUE}
miR2Target <- system.file("extdata", "miR2Target.csv", package="miRsponge")
miRTarget <- read.csv(miR2Target, header=TRUE, sep=",")
miRHomologyceRInt <- spongeMethod(miRTarget, method = "miRHomology")
spongenetwork_Cluster <- netModule(miRHomologyceRInt[, 1:2], modulesize = 2)
sponge_Module_DEA <- moduleDEA(spongenetwork_Cluster)
sponge_Module_FEA <- moduleFEA(spongenetwork_Cluster)
```

# Survival analysis of miRNA sponge modules
To make survival analysis of miRNA sponge modules, we implement 
'moduleSurvival' function. 

```{r}
miR2Target <- system.file("extdata", "miR2Target.csv", package="miRsponge")
miRTarget <- read.csv(miR2Target, header=TRUE, sep=",")
ExpDatacsv <- system.file("extdata", "ExpData.csv", package="miRsponge")
ExpData <- read.csv(ExpDatacsv, header=FALSE, sep=",")
SurvDatacsv <- system.file("extdata", "SurvData.csv", package="miRsponge")
SurvData <- read.csv(SurvDatacsv, header=TRUE, sep=",")
pcceRInt <- spongeMethod(miRTarget, ExpData, method = "pc")
spongenetwork_Cluster <- netModule(pcceRInt[, 1:2], modulesize = 2)
sponge_Module_Survival <- moduleSurvival(spongenetwork_Cluster, 
    ExpData, SurvData, devidePercentage=.5)
sponge_Module_Survival
```

Parameter 'devidePercentage' is used to set the percentage of high risk group. 

# Conclusions
miRsponge provides several functions to study miRNA sponge (also called ceRNA or miRNA decoy), including popular methods for identifying miRNA sponge interactions, and the integrative method to integrate miRNA sponge interactions from different methods, as well as the functions to validate miRNA sponge interactions, and infer miRNA sponge modules, conduct enrichment analysis of modules, and conduct survival analysis of modules. It could provide a useful tool for the research of miRNA sponges.

# References
[1] Ambros V. microRNAs: tiny regulators with great potential. Cell, 2001, 107:823–6.

[2] Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell, 2004, 116:281–97.

[3] Du T, Zamore PD. Beginning to understand microRNA function. Cell Research, 2007, 17:661–3.

[4] Esquela-Kerscher A, Slack FJ. Oncomirs—microRNAs with a role in cancer. 
Nature Reviews Cancer, 2006, 6:259–69.

[5] Lin S, Gregory RI. MicroRNA biogenesis pathways in cancer. 
Nature Reviews Cancer, 2015, 15:321–33.

[6] Salmena L, Poliseno L, Tay Y, et al. A ceRNA hypothesis: the Rosetta Stone 
of a hidden RNA language? Cell, 2011, 146(3):353-8.

[7] Cesana M, Cacchiarelli D, Legnini I, et al. A long noncoding RNA 
controls muscle differentiation by functioning as a competing endogenous 
RNA. Cell, 2011, 147:358–69.

[8] Poliseno L, Salmena L, Zhang J, et al. A coding-independent function 
of gene and pseudogene mRNAs regulates tumour biology. Nature, 2010,
465:1033–8.

[9] Hansen TB, Jensen TI, Clausen BH, et al. Natural RNA circles function 
as efficient microRNA sponges. Nature, 2013, 495:384–8.

[10] Memczak S, Jens M, Elefsinioti A, et al. Circular RNAs are a large 
class of animal RNAs with regulatory potency. Nature, 2013, 495:333–8. 

[11] Le TD, Zhang J, Liu L, et al. Computational methods for identifying 
miRNA sponge interactions. Brief Bioinform., 2017, 18(4):577-590.

[12] Tay Y, Rinn J, Pandolfi PP. The multilayered complexity of 
ceRNA crosstalk and competition. Nature, 2014, 505:344–52.

[13] Li JH, Liu S, Zhou H, et al. starBase v2.0: decoding miRNA-ceRNA, 
miRNA-ncRNA and protein-RNA interaction networks from large-scale 
CLIP-Seq data. Nucleic Acids Res., 2014, 42(Database issue):D92-7.

[14] Sarver AL, Subramanian S. Competing endogenous RNA database. 
Bioinformation, 2012, 8(15):731-3.

[15] Zhou X, Liu J, Wang W, Construction and investigation of 
breast-cancer-specific ceRNA network based on the mRNA and miRNA 
expression data. IET Syst Biol., 2014, 8(3):96-103.

[16] Xu J, Li Y, Lu J, et al. The mRNA related ceRNA-ceRNA landscape 
and significance across 20 major cancer types. Nucleic Acids Res., 
2015, 43(17):8169-82.

[17] Paci P, Colombo T, Farina L, Computational analysis identifies a 
sponge interaction network between long non-coding RNAs and messenger 
RNAs in human breast cancer. BMC Syst Biol., 2014, 8:83.

[18] Sumazin P, Yang X, Chiu HS, et al. An extensive microRNA-mediated 
network of RNA-RNA interactions regulates established oncogenic pathways 
in glioblastoma. Cell, 2011, 147(2):370-81.

[19] Tay Y, Kats L, Salmena L, et al. Coding-independent regulation of 
the tumor suppressor PTEN by competing endogenous mRNAs. Cell, 
2011, 147(2):344-57.

[20] Sardina DS, Alaimo S, Ferro A, et al. A novel computational method 
for inferring competing endogenous interactions. Brief Bioinform., 
2017, 18(6):1071-1081.

[21] Clauset A, Newman ME, Moore C. Finding community structure in 
very large networks. Phys Rev E Stat Nonlin Soft Matter Phys., 2004, 
70(6 Pt 2):066111.

[22] Enright AJ, Van Dongen S, Ouzounis CA. An efficient algorithm for 
large-scale detection of protein families. Nucleic Acids Res., 2002, 
30(7):1575-84.

[23] Kalinka AT, Tomancak P. linkcomm: an R package for the generation, 
visualization, and analysis of link communities in networks of 
arbitrary size and type. Bioinformatics, 2011, 27(14):2011-2.

[24] Bader GD, Hogue CW. An automated method for finding molecular 
complexes in large protein interaction networks. BMC Bioinformatics, 
2003, 4:2.

# Session information
```{r}
sessionInfo()
```