Table 1. Eigenvector centrality reconciles results for various co-evolution analyses.

For each pair of algorithms, non-parametric correlation coefficients (Spearman R²) are shown for pairwise co-evolution and eigenvector (EVC) scores, using either the unsubtracted (“Initial”) or subtracted (“Sub”) networks. As discussed in Methods, ZNDAMI is used in place of subtracted ZNMI.

Algorithms	Coefficient of determination (Spearman R2)
	Aldolase				LacI/GalR
	Pairwise		EVC		Pairwise		EVC
	Initial	Sub	Initial	Sub	Initial	Sub	Initial	Sub
ELSC vs McBASC	0.48	0.51	0.71	0.74	0.09	0.11	0.27	0.50
ELSC vs OMES	0.64	0.65	0.78	0.78	0.37	0.42	0.66	0.73
ELSC vs SCA	0.19	0.23	0.40	0.43	0.02	0.14	0.14	0.32
ELSC vs ZNMI	0.44	0.46	0.78	0.79	0.23	0.25	0.45	0.60
ELSC vs TEA-O				0.30				0.54
McBASC vs OMES	0.59	0.63	0.72	0.75	0.14	0.20	0.14	0.38
McBASC vs SCA	0.20	0.23	0.22	0.26	0.08	0.13	0.12	0.36
McBASC vs ZNMI	0.48	0.52	0.65	0.69	0.18	0.28	0.05	0.34
McBASC vs TEA-O				0.26				0.38
OMES vs SCA	0.25	0.25	0.31	0.31	0.31	0.34	0.43	0.55
OMES vs ZNMI	0.50	0.54	0.77	0.85	0.46	0.50	0.51	0.69
OMES vs tEA-O				0.26				0.70
SCA vs ZNMI	0.35	0.37	0.60	0.61	0.18	0.18	0.33	0.47
SCA vs TEA-O				0.25				0.42
ZNMI vs TEA-O				0.34				0.77
Median improvement		0.03	0.19	0.23 a		0.05	0.12	0.29a

^aMedian improvement does not include the TEA-O comparisons.