Probabilistic Inference of Biological Networks via Data Integration

Figure 3

Plot of the test accuracy ((a) -axis) and fraction of pairs predicted ((b) -axis) as a function of the -value cutoff (-axis) for (i) using all available pairwise and data kernels (, solid curve) and (ii) the top-performing pairwise kernel (, dashed curve). By increasing the -value cutoff, we increase the accuracy in our predictions but decrease the fraction of pairs for which we can make predictions.

(a)

(b)