ggraph package provides a ggplot-like grammar for plotting graphs and as such you
can produce very neat network visualizations. But as with ggplot, it takes a while to get
used to the grammar. There are already a few amazing Rstudio Addins that assist you with ggplot
(for example ggplotAssist and
but there has not been any equivalent tools for
ggraph. Till now. This post introduces
an Rstudio Addin which provides a tiny GUI for visualizing and analysing networks.
Installation and Usage
You can install the developer version with:
In order to work properly, the Package also needs the
smglr Package, which adds a new
In order to use the Addin, simply highlight a network in your script and select
snahelper from the Addin dropdown menu.
The GUI has the following components:
- Create and tweak a layout for your network
- Check and add note attributes
- Create the node styles
- Review edge attributes
- Create the edge styles
- Review Result
- Get the Code
Let’s go through them step-by-step.
The layout tab allows you to choose from all implemented algorithms in
the stress based layout from
smglr, which is the default and recommended choice.
See this blog post for an
explanation. In the tweak section you can move individual nodes around. The implementation
is still very crude, so don’t expect miracles from that. The layout is saved as node attributes
x and y.
Node Attribute Manager
The Node Attribute Manager shows all existing node attributes in a sortable table.
In addition, you can calculate some new ones (centrality and clustering).
NOTE: the centrality indices only work for undirected and unweighted networks so far.
This is where you can style your nodes. You can either do it manually, by choosing a color/size for all nodes together, or based on an attribute.
Edge Attribute Manager
So far, the Edge Attribute Manager only shows existing edge attributes. Future releases may include the possibility to calculate new ones.
You can style your edges here.
snahelper automatically detects if your network
is directed and adds arrows if the network is directed. The other options are similar to the nodes tab.
The curvature option allows you to bend edges. I highly suggest, that you only bend your
edges if there are two edges running between the same set of nodes to make them visible.
If you do not have reciprocal edges, then just leave it at 0.
The result tab shows the network in its full size. If you are satisfied with the results, hit the Done button and the R code to produce the plot is automatically inserted in your script.
So far I only incorporated what I consider the bare minimum to produce a good visualization of a network. Mostly because I did not want to overload the GUI. If you are missing something essential, open an issue and I will try to add the feature.