Handling and Analyzing Networks

In NetWalker, network views are handled within NetView windows. The NetView toolbar has several functions for visual handling and analyses of networks.

You can zoom in and out (scroll button) and drag the network (holding the right mouse button down) with your mouse. The different interaction colors correspond to different interaction types, as indicated in the coloring panel. You can change these colors from this panel. You can also filter these interactions only to display certain types of interactions from the filtering drop down menu ().

To demonstrate different functionalities in NetView, we will use the network we generated in the end of the section on NetWalk and EF Tables.

  1. First, in order to separate following analyses from the network we generated from the EF Table, select all nodes in the network, and copy to a new NetView window by clicking on Export button on the NetView toolbar () and select “to new Network”. This will generate a copy of the current network. Use this new copied network for following analyses.
  2. Try the different layouts in the drop-down box that has “Organic layout” as its current selection. Selecting any of the options in this box will redraw the network according to the selected layout algorithm. These different layout algorithms can reveal different characteristics of the network, such as its modular structure and inter-modular connectivity or hierarchical organization.
  3. Now we will perform a functional analysis of the network. Select all nodes, and click on “Func.Table” button to generate a FunTable containing functional enrichment analysis of the network selection.

    This is a FunTable containing enrichment scores of each functional category found in the selected network calculated by hypergeometric distribution function. However, unlike a FunTable generated from FunWalk, functional terms in this FunTable are mapped to genes, rather than interactions, and the last column shows genes in the selected network that belong to the given functional term.
  4. To rank functional terms by their p-values, click on the header of the column “Hyper-geometric p-value” to sort rows in the ascending order of p-value. Notice that rows with just 1 occurrence give a p-value of 0 (expected), but these are not of interest.
  5. Filter rows by “Number of occurrences” so that only rows where “Number of occurrences” column is higher than 5 is shown, and design another filter to include only rows with “Tree Level” attribute >4 (use “AND” clause for composite filter).
  6. Select functional terms (“cellular amine metabolic process”, “interphase of mitotic cell cycle” and “hexose biosynthetic process”) as shown below, all of which have significant p-values, and copy them to the clipboard (“Ctrl + C”).
  7. Go to the network view and paste these terms “Ctrl + V”. Redraw the network using “orthogonal layout”. You should see this:

    In this network, the functional groups were rendered as “ellipse” and the label font was changed. You can do this by selecting the three group nodes and altering their properties in the Properties window.
    One can also generate manual groupings of nodes by selecting a set of nodes and clicking on “Group”. The “AutoGroup” function will bring a dialog containing all functional terms contained in the current network, out of which you can select your desired functional terms to group the nodes in the network. However, for automated grouping of nodes, we recommend using the method we described here (i.e. generate a Func.Table, select desired functional terms from the FunTable and group by copy/paste).
  8. In network analyses, it may be desirable to see the expression levels of genes contained in the network. This can be done by filtering of dataset tables in TableView by copying a selected group of nodes in the network and pasting them onto the TableView window where the dataset of interest is. To do this:
         a.  Go to the TableView containing the original Doxorubicin_MCF7.txt dataset, select the “Names” column, and click on “Row names”. This will tell the TableView to use this column to match rows to genes in the Knowedgebase. Now, when you select rows in this TableView (after toggling row-selection), you will see details of corresponding gene annotations in the Details Window.
         b.  Go to the network, select all nodes, and do “Ctrl + C” to copy these nodes to the clipboard.
         c.  Go back to the TableView and paste “Ctrl + V”. In the dialog that comes up, select “Fitler rows” and press ok.
         d.  Now the dataset in TableView only shows rows corresponding to the genes selected in the network. You can analyze expression levels of these genes in the dataset by clustering as we did before (see Clustering and heatmaps).

    We would like to note that, instead of using copy/paste, the same filtering of the dataset could have been done by dragging the “Network from EF Table heatmap copy” node in the Object Tree window and dropping onto the TableView containing the Doxorubicin_MCF7.txt dataset.

  9. Next, we will generate a figure image out of our generated network. For this, click on “Export” button to bring the drop-down menu, and select either “as Image” or “as Image to Clipboard”. The former will save the current network view as an image file, the latter will copy the current network view to the clipboard as an image, which then can be pasted into any image processing software (Adobe Illustrator, Photoshop, Microsoft PowerPoint).

In this tutorial section, you learned to visually handle networks and perform various operations on networks for visual enhancements and functional analyses.

Now, you are ready to save the whole workspace containing the original dataset, heatmap derived from the dataset, several network objects, two EF Tables and three FunTable objects. Click on “Save Workspace” button () on the toolbar, or “Save workspace” under File menu. Name the project (e.g. “Doxorubicin”). The workspace will be saved into a folder “Doxorubicin”. This project can be later re-opened by “Load Workspace” in File menu, and selecting the “Doxorubicin” folder (it will be recognized the program as a NetWalker Workspace folder).

All of the analyses performed in this tutorial series and associated data and network objects have been saved and are available as a NetWalker Workspace folder in the Downloads section on this web site.