Figure 1From: New methods for separating causes from effects in genomics dataIllustration of the statistical testing procedure to assess significance of the causal orientation method performance (AUC/accuracy). a) AUC is computed using real data; b) AUC is computed using random data from the Normal distribution (null distribution) for the same gold standard as used with the real data, and this step is repeated 1,000 times; c) a p-value is calculated by comparing AUCs from the null distribution to the AUC obtained in the real data.Back to article page