Visualization and Data Exportation

To enable users to visualize the inferred network in graphical form, XMRF includes three plotting functions with slight variations to serve different purposes. First, the default plot function of the GMS class will draw the optimal inferred network and save it to a PDF file with the following command:

> plot(lpgm.fit, fn="lpgm.fit.net.png")

Second, the plotNet function allows users to plot a specific network with specific layout. For example, to plot the simulated network and the inferred network from section 4.1 with the same layout, the following commands can be used (Figure S7):

	> ml = plotNet(sim$B)
	> ml = plot(lpgm.fit, mylayout=ml)

Figure S7: Simulated network from XMRF.Sim(...,model="LPGM") (A) and inferred network estimated via XMRF(...,method="LPGM") with network sparsity determined via stability selection (B) plotted with the same layout.

The third plot function allows users to view the inferred network in other graph visualizing software such as the Cytoscape. The plotGML function will write the network in the graph modeling language (GML) format which then can be imported to Cytoscape. For example, with the following command:

	> plotGML(brca.lpgm, fn="brca.dnet.gml", weight=TRUE, vars=rownames(brca))

the estimated BRCA network (stored in brca.lpgm) was plotted and viewed from Cytospace (Figure S8). In this example, the width of the edges reflects the stability of the inferred edge.

Figure S8: The inferred relationships between cancer census genes from RNASeq data of BRCA patients from TCGA.

The inferred network in Figre S8 includes multiple associations reported in published literature, such as the associations of FOXA1, CCND1, and PBX1 with GATA3, link between ERBB2 and CDK12, and others. These results validate the usefulness of the implemented algorithms in the package.