Skip to main content
Figure 1 | BMC Bioinformatics

Figure 1

From: Dynamic probabilistic threshold networks to infer signaling pathways from time-course perturbation data

Figure 1

Evaluation on simulated data. A: The panel shows network topologies used to simulate data. Simulated Network 1 (SN1) is a simple feedforward network, whereas Simulated Network 2 (SN2) contains a negative feedback loop. B: Network reconstruction was performed for SN1, changing the number of unobserved proteins. Shown is the distribution of the area under the ROC curve (AUC ROC ) of 100 replicates of simulated data sets, over the fraction of unobserved proteins. From left to right: D-PBTN, DEPN, BDAGL, PBTN. The dashed line at AUC ROC  = 0.5 shows expected results for random guessing. C: The panel shows the distribution of AUC ROC values obtained for different levels of noise on SN1. Noise is introduced by switching the state of the indicated fraction of proteins in the "measured" data, thus introducing errors in the data. D: This panel shows performance of D-PBTN, DEPN, BDAGL AND PBTN on networks sampled from KEGG, with data simulated as described in methods. Shown are AUC ROC values of100 simulated data sets, generated from ten different KEGG subnetworks.

Back to article page