Table 3. Comparative analysis of GRACE’s predictions (ensemble model) vs all other methods for each independent validation (Suba, Aracyc) dataset for A. thaliana.

Methods/Datasets	Suba	Aracyc
GRACE vs GENIE3	FC = 3.7 (p = 5.4e–322)	FC = 9.8 (p = 6.3e–06)
GRACE vs CLR	FC = 3.3 (p = 2.746077e–187)	FC = 0.9 (p = 0.5)
GRACE vs GGM	FC = 2.3 (p = 2.9e–82)	FC = 4.0 (p = 0.03)
GRACE vs wGLASSO	FC = 7.7 (p = 1.9e–17)	—
GRACE vs iRafNet	FC = 2.2 (p = 8.0e–152)	FC = 5.2 (p = 0.015)

Fold changes (FC) and p-values based on Fisher’s exact test. ‘—’ indicates zero gold standard recovery rates of the compared method.