Advertisement

< Back to Article

Identifying Network Perturbation in Cancer

Fig 2

(A) Average receiver operating characteristic (ROC) curves from the experiments on synthetic data. We compare DISCERN with 7 alternative methods: 3 existing methods—LNS [35], D-score [36], and PLSNet [34]—and 4 methods we developed for comparison—pLNS, pD-score, D0 and pD0. (B) Comparison of the runtime (hours) between PLSNet and DISCERN for varying numbers of variables (p). The triangles mean the measured run times over specific values of p, and lines connect these measured run times. PLSNet uses the empirical p-values from permutation tests as scores, and DISCERN does not. For a large value of p, DISCERN is two to three orders of magnitude faster than PLSNet.

Fig 2

doi: https://doi.org/10.1371/journal.pcbi.1004888.g002