Skip to: Site menu | Main content

SpotXplore

SpotXplore is a plug-in for Cytoscape which enhances the visual analysis of gene expression obtained by, e.g., microarrays or the newer RNA-seq techniques in various ways. It enables visualization of multiple conditions or time series by glyphs that encode both expression value and statistical confidence. This visual mapping was introduced in GENeVis, which we have developed as a stand-alone application. SpotXplore can also detect subnetworks of interest based on the expression data, and it provides visualization techniques to explore these subnetworks.

SpotXplore features:

SpotXplore applied to a data set of the bacterium Bacillus subtilis, consisting of a gene regulatory network and gene expression measurements for four time points [larger image].

Publication

M. A. Westenberg, J. B. T. M. Roerdink, O. P. Kuipers, S. A. F. T. van Hijum. SpotXplore: a Cytoscape plugin for visual exploration of hotspot expression in gene regulatory networks. Bioinformatics, 26(22):2922-2923, 2010.


Download and installation

  1. Download and install Cytoscape [Cytoscape web site].
  2. Download SpotXplore archive (version 2010/08/04) [zip]. Install by unpacking the archive in the Cytoscape plugins folder. There is only a single Java jar file contained in the archive. Note: some browsers already unpack the archive, but do not rename the resulting file. If the plugin does not show up in Cytoscape's menu, try to rename SpotXplore.zip to SpotXplore.jar, and restart Cytoscape.
  3. Download an example data set [Cytoscape session file]. Details about the data set are in de user manual.
  4. Download the user manual [pdf].

Quick start: Run Cytoscape, load the session bsubtsession.cys, and invoke SpotXplore from the Plugins menu.


Time series visualization

Each time point is drawn as a rectangular glyph. The expression value determines the color, and the height can be dependent on significance (left) or expression value (right). Scaling by expression value gives the impression that the gene sspJ is strongly upregulated. This turns out to be noise, however, as it's p-value is 1.0. Scaling by significance (as shown on the left) conveys this directly. It also draws attention to significant changes in gene expression, which are not necessarily very large. [high resolution view]


Hotspot detection and visualization

SpotXplore supports a number of ways to find or define hotspots of gene expression:

  1. GiGA: automatic detection by graph-based iterative group analysis, see Breitling et al. (2004).
  2. Luscombe: trace-back algorithm by Luscombe et al. (2004) to identify subnetworks that are active.
  3. Manual: converts a selection set in Cytoscape to a hotspot. It is possible to construct multiple hotspots per time point or condition.

A gene regulatory network of B. subtilis superimposed with DNA microarray ratio data of B. subtilis wild-type over its ccpA deletion mutant. The time series consists of four time points corresponding to different phases of growth: (0) the early exponential phase (the onset of fast cell growth), (1) mid-exponential phase (fast cell growth), (2) end-exponential phase (nutrients start slightly limiting the growth), and (3) the stationary phase of growth (no growth of cells and start of cell death). The image shows the early exponential growth phase (time point 0). Hotspot are detected by the GiGA method with the group size parameter set to 25 and the p-value threshold to 1.0E-10. Nodes and edges which are member of a hotspot are drawn opaque, whereas the remaining nodes and edges are drawn translucent to provide a context.

[Play an animation of the time series].
The short 'movie of life' makes apparent that the impact of the ccpA mutation dynamically develops and intensifies during growth of B. subtilis cells. At the last time point, one hotspot is detected which overlaps almost completely with the SigB regulon. This regulon is involved in the response to harmful environmental conditions, such as heat, osmotic, acid, or alkaline shock. This hotspot is interesting, since it explicitly reveals that the SigB regulon is recruited stronger during the late growth stages of the wild-type strain than the ccpA deletion strain. Additional data might be required to provide a biological explanation for this effect.