Skip to main content
. 2007 Dec 12;104(51):20449–20453. doi: 10.1073/pnas.0706339104

Fig. 4.

Fig. 4.

Evaluation of the probabilistically inferred human bZIP interactions with various levels of noise applied to the input interaction scores. (a) No noise added. (b) Adding Gaussian noise with a SD of 10. (c) Adding Gaussian noise with a SD of 20. The phylogenetic tree for the human bZIPs is derived from the LZ regions of the bZIP proteins (see Methods). An interaction is represented by a filled cell in the matrix; within-family interactions are colored. (Lower) Input human interaction data. (Upper) Interactions predicted at >50% probability by inference over the full evolutionary model. ROC plots show the quality of each input (dashed gray line) and output (black line) dataset by comparison to a subset of human bZIP pairs with unambiguous experimental interaction data. Interaction matrices were prepared by using TVi (31).