How do we know that a network analysis method works? Testing against synthetic data (where the truth is known by construction) is a common approach, but lacks the statistical idiosyncrasies of experimental data. An alternative approach is to test whether the method yields consistent results across multiple studies of the same phenotype.
In this paper, we describe a framework for evaluating the reliability of network analysis methods using a curated suite of cancer datasets, and apply it to evaluate methods available on BioConductor.
