β versions of GGobi & rggobi

This page describes how to test out out the development version of GGobi and rggobi.

Installation

The installation process is platform-dependent. Instructions are given for installing both GGobi and rggobi. If you want to experiment with the new accessibility and extensibility of GGobi from R you'll need to install both. (Note that rggobi depends on the RGtk2 package; it should be automatically installed with rggobi.)

Windows

Linux

Mac

Coming soon!

Notes on the beta version

You'll notice that the new GGobi has (a few) less features than the released version and that it is a lot slower and buggier. This is because:

Rggobi now has a direct binding to the new GGobi objects, allowing the extension of GGobi classes from R. Most of GGobi remains to be refactored into objects, so this functionality is limited, but you can already manipulate the data pipeline.

Toying around: A demonstration

An application of this to graph layout is shown in the movie above. A pipeline stage has been written in R that generates a graph layout using the IPSep-CoLa algorithm in the adaptagrams library. IPSep-CoLa operates similar to multi-dimensional scaling (MDS), except it applies "separation constraints" to the layout, that enforce ordering and separation between pairs of nodes. In the movie, when a user brushes a node in the graph, a constraint is added to the IPSep-CoLa algorithm which forces the brushed (purple) points to lie to the right of the yellow points.

To try this demo on your own machine, install the rcola package:

install.packages("rcola", repos="http://www.ggobi.org/r")

which includes adaptagrams and binds it to R. Rcola requires the Bioconductor graph package. Rcola does not work very well on Windows, so expect crashes. It works just fine on Mac and Linux. After installing rcola, it is possible to run the demo on your own machine by entering demo(rcola) after loading the rggobi package in R. The source code is fairly well commented, so it can be read to gain an understanding of how the demo works. A deeper understanding will require reading the documentation of the GGobi API.

Documentation