Sparse and Compositionally Robust Inference of Microbial Ecological Networks
Fig 4
Precision-recall performance on synthetic datasets.
a) Red = S-E(glasso), orange = S-E(MB), purple = SparCC, blue = CCREPE, green = Pearson correlation, black = random. Area under precision-recall (AUPR) vs. number of samples n for different κ values are depicted. Bars represent average over 20 synthetic datasets, and error bars represent standard error. Asterisks denote conditions under which SPIEC-EASI methods had significantly higher AUPR relative to all other control methods (P<0.05 for all one-sided T tests). b) Representative precision-recall curves for p = 68, n = 102, κ = 100; solid and dashed lines denote SPIEC-EASI and control methods, respectively.