MIENTURNET: an interactive web tool for microRNA-target enrichment and network-based analysis

Background miRNAs regulate the expression of several genes with one miRNA able to target multiple genes and with one gene able to be simultaneously targeted by more than one miRNA. Therefore, it has become indispensable to shorten the long list of miRNA-target interactions to put in the spotlight in order to gain insight into understanding the regulatory mechanism orchestrated by miRNAs in various cellular processes. A reasonable solution is certainly to prioritize miRNA-target interactions to maximize the effectiveness of the downstream analysis. Results We propose a new and easy-to-use web tool MIENTURNET (MicroRNA ENrichment TURned NETwork) that receives in input a list of miRNAs or mRNAs and tackles the problem of prioritizing miRNA-target interactions by performing a statistical analysis followed by a fully featured network-based visualization and analysis. The statistics is used to assess the significance of an over-representation of miRNA-target interactions and then MIENTURNET filters based on the statistical significance associated with each miRNA-target interaction. In addition, the holistic approach of the network theory is used to infer possible evidences of miRNA regulation by capturing emergent properties of the miRNA-target regulatory network that would be not evident through a pairwise analysis of the individual components. Conclusion MIENTURNET offers the possibility to consistently perform both statistical and network-based analyses by using only a single tool leading to a more effective prioritization of the miRNA-target interactions. This has the potential to avoid researchers without computational and informatics skills to navigate multiple websites and thus to independently investigate miRNA activity in every cellular process of interest in an easy and at the same time exhaustive way thanks to the intuitive web interface. The web application along with a well-documented and comprehensive user guide are freely available at http://userver.bio.uniroma1.it/apps/mienturnet/ without any login requirement.


Tutorial
From the Home page of MIENTURNET, click on Data Upload link on the top navigation bar to start the analysis.

Entering a list of miRNAs Data Upload link
If the user is interested to find targets of a given list of miRNAs, from the Data Upload page, select miRBase ID as type of the input list.

Input list identifiers
The input list must include identifiers of the mature miRNAs as miRBase ID (e.g., hsa-miR-15a-5p). The tool processes the input file and immediately provides in the IDs found collapsible panel the total number of miRNAs recognized by miRBase database together with a table that includes: microRNA names, miRBase accession IDs as hyperlinks to miRBase, and microRNA mature sequences.
In the input list table, the user has also the possibility to: i.
visualize the first 10, 25, 50, 100 entries by clicking the drop-down menu at the bottom-left ii.
navigate the table trough the navigation controller buttons at the bottom-right iii.
search for a specific miRNA in the input list by typing the name in the search bar at the topright. iv.
copy the table to the clipboard, export it in a comma separated format (CSV) or in EXCEL format through the dedicated buttons placed at the top-left.
The list of not recognized miRNAs (if any) can be viewed by clicking on the IDs not found collapsible panel.
An error message appears either if the input list contains all miRBase IDs that are not valid or if the organism is incorrectly selected.
For each gene in the chosen database (TargetScan or miRTarBase), the hypergeometric test was used to calculate the significance (p-value <0.05) of the enrichment of the input list in miRNAs targeting that gene.
The p-value is computed as where M is the dimension of the universe, that is the number of all predicted (validated) miRNA-target interactions encompassed in TargetScan (miRTarBase); K is the number of predicted (validated) miRNA-target interactions encompassed in TargetScan (miRTarBase) for the selected target gene; N is the number of miRNA families (number of miRNAs) of the input list recognized by TargetScan (miRTarBase); X is the number of miRNA families (number of miRNAs) for which predicted (validated) miRNA-target interactions, for the selected gene, exist.
The results of the enrichment analysis are presented as 1 : i.
a bar plot where the Y-axis refers to the top ten target genes resulted from the enrichment analysis, while the X-axis represents the number of miRNAs targeting them. The color code reflects the FDR value increasing from red to blue. This plot is resizable by grabbing and dragging an active corner at the bottom right of the page.
ii .  an interactive table where the user has also always the possibility to:   1. visualize the first 10, 25, 50, 100 entries by clicking the drop-down menu at the bottomleft  2. navigate the table trough the navigation controller buttons at the bottom-right  3. search for a miRNA or a Target Gene of interest by typing the name in the search bar at  the top-right  4. copy the table to  In order to build the network of miRNA-target interactions identified by the enrichment analysis, click on Network menu and choose the reference database (i.e., TargetScan and/or miRTarBase) from the drop-down menu.

Selecting miRNA-target interactions database and filtering
By choosing TargetScan, the results can be filtered by using a threshold on: i. the minimum number of miRNA-target interactions ii.
the adjusted p-values obtained from the miRNA-target enrichment analysis iii.
the probability of conserved targeting (P CT ) parameter [5] provided by TargetScan  13 The user can choose to apply one of these filters, more than one or none. By default, the first two thresholds are set to the value chosen in the miRNA-target Enrichment page. The other two parameters, CWCS and P CT , are set by default to 0.9 and 0.5, respectively.
By choosing miRTarBase, the results can be filtered according to i. the minimum number of miRNA-target interactions ii.
the adjusted p-values obtained from the miRNA-target enrichment analysis iii.
the type of Evidence categories used by miRTarBase to validate the miRNA-target interactions: a. 'Strong' for considering strong experimental methods (e.g., Luciferase assay, Western) b. 'Weak' for considering weaker experimental evidence (e.g., CLIP) c. 'Strong and Weak' for considering both strong and weak experimental methods As before, by default the first two thresholds are set to the value chosen in the miRNA-target Enrichment page. Instead, the Evidence categories filter is configured as 'Strong'.
Caveat: regardless of the database choice, the user that is not interested in the enrichment analysis can skip the previous step (i.e., the user can avoid to click on miRNA-target Enrichment link) and jump to the next step of Network menu. In this case, MIENTURNET will still perform the enrichment analysis, but will set the default value for the adjusted p-values to 1.

Network visualization and analysis
Both for TargetScan and miRTarBase, the user can interactively play with the miRNA-interaction network by pressing the following tabs: 1. Network tab: it allows to visualize, customize, and download the network in the form of an edgelist (i.e., a 2-column table where each row represents a single edge) 2. Network properties tab: it allows to perform a network analysis 3. miRNA-target Table tab: it allows to view and download a table where the columns are the miRNAs with the corresponding own targets In the following, there is a detailed description for each of these tabs 3 .

Network tab
From Network tab, the user can visualize the miRNA-target interaction network: i. as an image that can be interactively explored and customized ii.
as a 2-column table in the form of an edgelist (i.e., each row represents a miRNA-target interaction).
 the network layout (details about the layout algorithms are available at http://igraph.org)  the size of source/target nodes  the shape of source/target nodes  the color of source/target nodes. By default, source nodes (miRNA families or single miRNAs) are blue circles, while target nodes (mRNAs) are yellow circles.
On the network image, the user has also the possibility to: i. move the network around by using the up-down left-right arrows ii.
zoom in/out the image by using the plus/minus buttons iii. select a specific node by scrolling down the drop-down menu "Select by id" at the top-left iv. select a specific node by placing the pointer over the node to be selected. Note, the selection will interest the node and its first neighbors. For example, if a miRNA is selected both miRNA and its targets will be highlighted. v.
move one or more node by selecting and dragging them with mouse The user can export the network picture as: i. a dynamic image in HTML format, by clicking first SAVE NETWORK button and then DOWNLOAD NETWORK button at the top-left ii. a static image in PNG format, by pressing down on the right mouse button (i.e., rightclicking) i. visualize the first 10, 25, 50, 100 entries by clicking the drop-down menu at the bottom-left ii.
navigate the table trough the navigation controller buttons at the bottom-right  iii. search for a miRNA or a target gene of interest by typing the name in the search bar at the topright iv.
copy the table to the clipboard, export it in a comma separated format (CSV) or in EXCEL format through the dedicated buttons placed at the top-left.
One or more miRNAs in the input list, although recognized in miRBase, could not be in the chosen database (TargetScan or miRTarBase). In this case, the miRNAs not found are listed in the IDs not found collapsible panel.

Network Properties tab
From Network Properties tab, the user can: i. visualize a table with the topological properties for each node 4 , where it is possible to: 1. sort each column in decreasing or increasing order by clicking on the up-down arrows close to the column names 2. visualize the first 10, 25, 50, 100 entries by clicking the drop-down menu at the bottomleft 3. navigate the table trough the navigation controller buttons at the bottom-right 4. search for a node of interest by typing the name in the search bar at the top-right 5. copy the table to the clipboard, export it in a comma separated format (CSV) or in EXCEL format through the dedicated buttons placed at the top-left In particular, the columns of the table report:  degree -the number of incoming and outgoing edges of each node  betweennessthe number of shortest paths 5 going through the node  average shortest path lengththe average length of the shortest paths between a node and all other nodes in the network  eccentricitythe shortest path distance of a node from the farthest other node in the network  clustering coefficientmeasurement the probability that the adjacent nodes of a node are connected. The clustering coefficient is set to 0 for nodes with degree zero and one.  closenessthe inverse of the average length of the shortest paths between a node and all other nodes in the network. Note that the closeness centrality measure is not well-defined for disconnected graphs. If there is no (directed) path between node i and j, then the total number of nodes is used in the formula instead of the path length. ii.
visualize the following network plots: a. a bar plot of the mRNA degree, where X-axis refers to the first 30 target genes (sorted in a decreasing order according to the degree) and Y-axis refers to their degree (i.e., number of miRNA targeting them) b. a bar plot of the miRNA degree, where X-axis refers to the first 30 miRNAs (sorted in a decreasing order according to the degree) and Y-axis refers to their degree (i.e., number of their targets) c. the nodes degree distribution ( ), with being the degree, by plotting log on the Xaxis and log ( ) on the Y-axis. Then, the method of least squares is used to find the best-fitting curve log~log , where α is the constant to estimate. The sum of the squared residuals 2 (i.e., the sum of the vertical deviations from the best-fitting line) is reported together with the estimation of the coefficient α.
All plots can be exported as a compressed .zip file by clicking on SAVE DEGREE PLOTS button.
From miRNA-target

Functional Enrichment link
In order to perform the functional enrichment analysis of target genes of selected miRNAs, click on Functional Enrichment link and choose TargetScan or miRTarBase 6 .
Currently, MIENTURNET allows to query the following annotation databases: KEGG, REACTOME, WikiPathways, Disease Ontology. Note that the last one will be available only when Homo sapiens is the selected organism.
The user can select one or more miRNA families (or single miRNA in the case of miRTarBase) by clicking on each item or by typing the name in the search box. In this example, the first three miRNA families have been selected.
14 The selected miRNA families will appear in a table on the right along with the number of their interactions.
To start the analysis, click the ENRICH button.
Caveat: The function enrichment analysis will regard the targets of each selected miRNA, separately. The user could see a progress bar in action at the bottom-right of the page.

1.
an interactive table reporting: the miRNA name, the category name of the annotation database (e.g., KEGG ID), the category description, the p-value and the adjusted p-value (FDR), the list of target genes, the number of targets genes. 2.
a dot plot, where the Y-axis reports the annotation categories (e.g., KEGG pathways) and the X-axis reports the miRNA family with the number of recognized targets in round brackets. The colors of the dots represent the adjusted p-values (FDR), whereas the size of the dots represents gene ratio (i.e., the number of miRNA targets found annotated in each category over the number of total genes annotated in that category). All the plots are resizable by grabbing and dragging an active corner at the bottom right of the page. 24

Entering a list of mRNAs Data Upload link
If the user is interested to find miRNAs targeting a given list of mRNAs, from the Data Upload page, select Official Gene Symbol as type of the input list.
Select the organism from the drop-down menu.
Enter the input list by using one of the following options: 1) Upload a file if the list is saved in a text file on the own computer and then click BROWSE to quickly access to the file 2) Paste your list in the specific text area and then click SUBMIT 3) Load an example list for playing with data. With this choice, the user has the possibility to explore the functionality of MIENTURNET by exploiting a usage example.
Hereafter, we will show how MIENTURNET works by loading the example input list of mRNAs selected by the authors. Note that this list represents the differentially expressed genes resulting from a recent computational study of glioblastoma cells that was aimed to identify genes playing a key role in the transition from stem-like tumor propagating cells towards differentiated glioblastoma cells [6].

Input list identifiers
The input list must include identifiers of mRNAs as Official Gene Symbol (e.g., PTEN for human species, Pten for mouse species). The tool processes the input file and immediately provides in the IDs found collapsible panel the total number of mRNAs recognized by NCBI database together with a table that includes: gene symbols of the input list, the corresponding updated gene symbols (i.e., the latest version of the gene name according to NCBI database), Entrez IDs as hyperlinks to NCBI Gene, and gene descriptions.
Caveat: the analysis will proceed by considering the updated gene symbols.
In the input list table, the user has also the possibility to: i. visualize the first 10, 25, 50, 100 entries by clicking the drop-down menu at the bottom-left ii.
navigate the table trough the navigation controller buttons at the bottom-right iii.
search for a specific miRNA in the input list by typing the name in the search bar at the topright. iv.
copy the table to the clipboard, export it in a comma separated format (CSV) or in EXCEL format through the dedicated buttons placed at the top-left.
The list of not recognized mRNAs (if any) can be viewed by clicking on the IDs not found collapsible panel.
An error message appears either if the input list contains Official Gene Symbols that are not valid, or if the organism is incorrectly selected.

miRNA-target Enrichment link
By clicking on miRNA-target Enrichment link, MIENTURNET will perform a statistical analysis for over-representation of miRNA-target interactions in the input gene list.
The user has the possibility to perform the enrichment analysis by using a threshold on: i. the minimum number of miRNA-target interactions (default is 2) ii.
the adjusted p-values (default is 1) From miRNA-target Enrichment page, the user can choose the reference database for the miRNAtarget interactions (i.e., TargetScan and/or miRTarBase) to be used to perform the enrichment analysis.
Caveat: regardless of the database choice, MIENTURNET performs the miRNA-target enrichment analysis searching for both predicted and validated interactions. A progress bar in action will appear at the bottomright of the page as long as both analyses are not completed.
By clicking on TargetScan tab, the computationally predicted miRNA-target interactions will be considered.
Caveat: TargetScan predicts biological targets of miRNAs by searching for the presence of conserved sites that match the seed region (i.e., the region comprising nucleotides 2-7 at the 5′end of the mature miRNA sequence) of each miRNA [3]. Note that, a miRNA family is comprised of miRNAs with the same seed sequence. This implies that members of the same miRNA family all share the same predicted targets. As a consequence, in this case, the enrichment analysis was performed considering miRNA families instead of single miRNAs.
By clicking on miRTarBase tab, the experimentally validated miRNA-target interactions will be considered.
Caveat: miRTarBase collects miRNA-target interactions experimentally validated by reporter assay, western blot, microarrays, and next-generation sequencing experiments. As a consequence, in this case, the enrichment analysis was performed considering miRNAs instead of miRNA families.
For each miRNA in the chosen database (TargetScan or miRTarBase), the hypergeometric test was used to calculate the significance (p-value <0.05) of the enrichment of the input list in targets of that miRNA.
The p-value is computed as where M is the dimension of the universe, that is the number of all predicted (validated) miRNA-target interactions encompassed in TargetScan (miRTarBase); K is the number of predicted (validated) miRNA-target interactions encompassed in TargetScan (miRTarBase) for the selected miRNA family (selected miRNA); N is the number of genes of the input gene list recognized by TargetScan (miRTarBase); X is the number of genes for which predicted (validated) miRNA-target interactions, for the selected miRNA family (selected miRNA), exist.
The results of the enrichment analysis are presented as 8 : i.
a bar plot where the Y-axis reports the top ten miRNA families resulted from the enrichment analysis, while the X-axis reports the number of their targets. The color code reflects the FDR value increasing from red to blue. This plot is resizable by grabbing and dragging an active corner at the bottom right of the page.
ii. an interactive table where the user has also always the possibility to: 1. visualize the first 10, 25, 50, 100 entries by clicking the drop-down menu at the bottomleft  2. navigate the table trough the navigation controller buttons at the bottom-right  3 Note that, selecting miRTarBase, the first two columns of the above table will be replaced with a single column with the miRNA names (i.e., miRbase ID).
In order to build the network of miRNA-target interactions identified by the enrichment analysis, click on Network menu and choose the reference database (i.e., TargetScan and/or miRTarBase) from the drop-down menu.

Selecting miRNA-target interactions database and filtering
By choosing TargetScan, the results can be filtered by using a threshold on: i. the minimum number of miRNA-target interactions ii.
the adjusted p-values obtained from the miRNA-target enrichment analysis iii.
the probability of conserved targeting (P CT ) parameter [5] provided by TargetScan The user can choose to apply one of these filters, more than one or none. By default, the first two thresholds are set to the values chosen in the miRNA-target Enrichment page. The last two parameters, CWCS and P CT , are set by default to 0.9 and 0.5, respectively.
By choosing miRTarBase, the results can be filtered according to: i. a threshold on the minimum number of miRNA-target interactions ii.
a threshold on the adjusted p-values obtained from the miRNA-target enrichment analysis iii.
the type of Evidence categories used by miRTarBase to validate the miRNA-target interactions: a. 'Strong' for considering strong experimental methods (e.g., Luciferase assay, Western) b. 'Weak' for considering weaker experimental evidence (e.g., CLIP) c. 'Strong and Weak' for considering both strong and weak experimental methods The user can choose to apply one of these filters, more than one or none. As before, by default the first two thresholds are set to the values chosen in the miRNA-target Enrichment page. Instead, the Evidence categories filter is configured as 'Strong'.
Caveat: regardless of the database choice, the user that is not interested in the enrichment analysis can skip the previous step (i.e., the user can avoid to click on miRNA-target Enrichment link) and jump to the next step of Network menu. In this case, MIENTURNET will still perform the enrichment analysis, but will set the default value for the adjusted p-values to 1.

Network visualization and analysis
Both for TargetScan and miRTarBase, the user can interactively play with the miRNA-interaction network by pressing the following tabs: 1. Network tab: it allows to visualize, customize, and download the network in the form of an edgelist (i.e., a 2-column table where each row represents a single edge) 2. Network properties tab: it allows to perform a network analysis 3. miRNA-target Table tab: it allows to view and download a table where the columns are the  miRNAs with the corresponding own targets In the following, there is a detailed description for each of these tabs 10 .

Network tab
From Network tab, the user can visualize the miRNA-target interactions network: i. as an image that can be interactively explored and customized ii.
as a 2-column table in the form of an edgelist (i.e., each row represents a miRNA-target interaction).
 the network layout (details about the layout algorithms are available at http://igraph.org)  the size of source/target nodes  the shape of source/target nodes  the color of source/target nodes. By default, source nodes (miRNA families or single miRNAs) are blue circles, while target nodes (mRNAs) are yellow circles.
On the network image, the user has also the possibility to: i. move the network around by using the up-down left-right arrows ii.
zoom in/out the image by using the plus/minus buttons iii. select a specific node by scrolling down the drop-down menu "Select by id" at the top-left iv. select a specific node by placing the pointer over the node to be selected. Note, the selection will interest the node and its first neighbors. v.
move one or more node by selecting and dragging them with mouse The user can export the network picture as: i. a dynamic image in HTML format, by clicking first SAVE NETWORK button and then DOWNLOAD NETWORK button at the top-left ii. a static image in PNG format, by pressing down on the right mouse button (i.e., rightclicking) i. visualize the first 10, 25, 50, 100 entries by clicking the drop-down menu at the bottom-left  ii.  navigate the table trough the navigation controller buttons at the bottom-right  iii. search for a miRNA or a target gene of interest by typing the name in the search bar at the topright iv.
copy the table to the clipboard, export it in a comma separated format (CSV) or in EXCEL format through the dedicated buttons placed at the top-left.
One or more mRNAs in the input list, although recognized in NCBI, could not be in the chosen database (TargetScan or miRTarBase). In this case, the mRNAs not found are listed in the IDs not found collapsible panel.
ii. visualize the following network degree plots: a. a bar plot of the mRNA degree, where X-axis refers to the first 30 target genes (sorted in a decreasing order according to the degree) and Y-axis refers to their degree (i.e., number of miRNA targeting them) b. a bar plot of the miRNA degree, where X-axis refers to the first 30 miRNAs (sorted in a decreasing order according to the degree) and Y-axis refers to their degree (i.e., number of their targets) c. the nodes degree distribution ( ), with being the degree, by plotting log on the Xaxis and log ( ) on the Y-axis. Then, the method of least squares is used to find the best-fitting curve log~log , where α is the constant to estimate. The sum of the squared residuals 2 (i.e., the sum of the vertical deviations from the best-fitting line) is reported together with the estimation of the coefficient α.
All plots can be exported as a compressed .zip file by clicking on SAVE DEGREE PLOTS button.
From miRNA-target Table tab, the user can visualize an interactive table where the columns are the mRNAs of the input list found in the selected database (TargetScan or miRTarBase) with the corresponding miRNAs targeting them. The columns of the table are sorted in a decreasing order according to the number of miRNAs targeting each gene. The user has always the possibility to: i. visualize the first 10, 25, 50, 100 entries by clicking the drop-down menu at the top-left ii.
navigate the table trough the navigation controller buttons at bottom-right iii.
search for a miRNA or a Target Gene of interest by typing the name in the search bar at the top-right iv.
copy the table to the clipboard, export it in a comma separated format (CSV) or in EXCEL format through the dedicated buttons placed at the top-left.
In order to perform the functional enrichment analysis of target genes of selected miRNAs, click on Functional Enrichment link and choose TargetScan or miRTarBase 13 .
Currently, MIENTURNET allows to query the following annotation databases: KEGG, REACTOME, WikiPathways, Disease Ontology. Note that the last one will be available only when Homo sapiens is the selected organism.
Firstly, the user can select one or more miRNA families (or single miRNA in the case of miRTarBase) by clicking on each item or by typing the name in the search box. In this example, the first three miRNA families have been selected.
14 The selected miRNA families will appear in a table on the right along with the number of their interactions.
Secondly, the user can select the background set. Currently, the choice is between: "Whole Genome" (i.e., all annotated genes in the annotation database will be used as reference set), or "Your input list of genes" (i.e., the user's input list of gens will be used as reference set).
To start the analysis, click the ENRICH button.
Caveat: The function enrichment analysis will regard the targets of each selected miRNA, separately. The user could see a progress bar in action at the bottom-right of the page.