Table 2:

Analytically derived and simulated metrics for different designs.

Pairs	Design	$\mathit{M}\mathit{S}{\mathit{E}}_{\stackrel{\wedge }{\mathit{\Delta }\mathit{G}}}$	$\mathit{M}\mathit{S}{\mathit{E}}_{\mathit{\Delta }\stackrel{\wedge }{\mathit{\Delta }\mathit{G}}}$b	$\mathit{M}\mathit{S}{\mathit{E}}_{\mathit{\Delta }\stackrel{\wedge }{\mathit{\Delta }\mathit{G}}}^{\mathit{A}\mathit{l}\mathit{l}}$c	Median ρ d	Accuracye
20	Random*	4.08	0.95	3.79	0.60 (0.38–0.77)	30.5%
	D-optimal	3.83	0.95	3.50	0.60 (0.38–0.77)	31.5%
	A-optimal	1.42	0.95	1.83	0.70 (0.54–0.82)	35.3%
30	Random*	1.66	0.63	1.33	0.77 (0.62–0.87)	40.9%
	D-optimal	1.36	0.63	0.91	0.81 (0.68–0.88)	49.2%
	A-optimal	1.08	0.63	1.10	0.79 (0.67–0.88)	45.3%
50	Random*	0.98	0.38	0.54	0.87 (0.78–0.92)	55.0%
	D-optimal	0.92	0.38	0.44	0.89 (0.80–0.93)	58.3%
	A-optimal	0.79	0.38	0.50	0.88 (0.79–0.93)	57.5%

^*The averages across 5,000 different randomly selected designs.

^a$MS{E}_{\stackrel{\wedge }{\mathrm{\Delta }G}}$ was analytically derived from equation (3).

^b$MS{E}_{\mathrm{\Delta }\stackrel{\wedge }{\mathrm{\Delta }G}}$ was analytically derived from equation (3) and includes only pairwise differences with corresponding FEP values.

^c$MS{E}_{\mathrm{\Delta }\stackrel{\wedge }{\mathrm{\Delta }G}}^{All}$ was analytically derived from equation (3) and includes all _NC₂ = 190 possible pairwise differences.

^dSpearman’s rank correlation (ρ) between the ∆G^True and estimated $\stackrel{\wedge }{\mathrm{\Delta }G}$ values of 20 ligands with the 15th and 85th quantiles in parenthesis.

^eProbability of correctly identifying the ligand with the lowest absolute free energy.