Tutorial: Create & Filter the Weighted Network

This tutorial introduces VisANT’s basic capability for constructing and displaying networks with weighted edges; i.e. edges that reflect link reliability.

We use an Edge-List as the format for input data, The linked nodes are labeled with COG ids which are not organism specific. Therefore users should always select a specific organism before adding data.

Step 1: Prepare the data using Edge List

To create a network with weighted edges, we represent the network as a five column matrix, with the number of rows equal to the number of edges. As an example we will weight a network constructed by phylogenetic profiling. In each row the first two cols have the IDs of the paired orthologs, the third column has a 0 (in this case) since the connecting edge has no direction, the fourth column is the method ID (phylogenetic profiling), and the 5th column is the probability that the profile similarity could have occurred purely by chance. For this tutorial, we use following data:

COG0001 COG0007 0 M0037 0.427687442686038
COG0001 COG0045 0 M0037 0.323141599227162
COG0001 COG0054 0 M0037 0.317120174092673
COG0001 COG0063 0 M0037 0.262304516458299
COG0001 COG0074 0 M0037 0.323141599227162
COG0001 COG0082 0 M0037 0.35787446628169
COG0001 COG0108 0 M0037 0.317120174092673
COG0001 COG0111 0 M0037 0.262304516458299

COG0669 COG1158 0 M0037 0.390289978866581
COG0669 COG1159 0 M0037 0.605561256310388
COG0669 COG1162 0 M0037 0.290270944130449
COG2190 COG3414 0 M0037 0.403796627789251
COG2190 COG3444 0 M0037 0.337013380698925
COG2190 COG3623 0 M0037 0.305619736118715
COG2190 COG3711 0 M0037 0.403796627789251
COG2190 COG3715 0 M0037 0.337013380698925
COG2190 COG3716 0 M0037 0.337013380698925
COG2190 COG3835 0 M0037 0.275403852766849
COG2190 COG4668 0 M0037 0.492058472821917
COG2190 COG4987 0 M0037 0.270698940335453
COG2190 COG4988 0 M0037 0.294197134509517
COG2191 COG2266 0 M0037 0.257053580945912
COG2191 COG2450 0 M0037 0.317639621878071
COG2191 COG2511 0 M0037 0.257053580945912
COG2191 COG2811 0 M0037 0.434062434523347
COG2191 COG2888 0 M0037 0.257053580945912
COG2191 COG3390 0 M0037 0.304200087240157
COG2191 COG3620 0 M0037 0.304200087240157
COG2191 COG3635 0 M0037 0.318647443121954

Note: You can find a list of method ID in the method table which is available under View menu. In the same place you can create your own method if you can not find a suitable one for your data.

Note: For more information about data format, please reference here.

Note: If you have a lot of data, you can either copy/paste in several separate steps, or run VisANT as application and load the data from a file.

Step 2: Create the weighted network

  • Start VisANT by clicking here or on Start VisANT at the right-upper corner of VisANT banner. In this tutorial, we run VisANT as an applet. VisANT can run in different mode.

  • Clear the network panel by clicking the Clear button in VisANT's ToolBox (left side) (Fig. A)

  • Paste the data shown in step 1 into the text box as shown in Fig. B, and click the Add button (Fig. B).

  • Click the Fit To Page (Fig. C) to make the network fit to the view area.

The network will be created as shown in the figure below with edge's thickness associated with its weight. This association can be removed by deselecting the Fine Arts option (Fig. D).

 

Note: The Fine Arts will automatically be turned off when you click the Add button if the total number of edges and nodes is more than 12,000 to improve the performance.

Step 3: Filter the weighted network

  • VisANT can hide edges whose weight is less than a cutoff value that can be specified as shown in the left-upper corner in VisANT (Fig. A).

  • As an example, type in 0.32 and press the enter key. Those edges with weight less than 0.32 are suppressed as illustrated in following figure.
    Note: You can also use your mouse to click the up/down arrow to change the cutoff value (the up/down key also works)

  • The filtering only removes the edge, to remove the nodes that have no edges linking to them, invoke the degree distribution as shown in figure below.

  • Filtering only removes the edge, to remove the nodes that have no edges, invoke the degree distribution as shown in the figure below.

  • Change the cutoff value and observe the change in the network.

  • In the degree distribution plot, make sure that the option Link to Network (Fig. B) is selected. Drag a rectangle to include the point with degree equals zero (Fig. C).

  • The nodes with degree=0 will be selected as shown in the figure. To remove them, press Delete key or select Delete Selected Nodes under Edit menu.