Table 1.

Comparison of values for ΔG_mutate and ΔΔG_bind, in kJ·mol⁻¹.

Alchemical Mutation	ΔGmutate,solv	ΔGmutate,bound	ΔΔGbind,calc	ΔΔGbind,exp
A → G	−3.9 ± 0.4	6.8 ± 0.7
A ← G	4.2 ± 0.4	−1.6 ± 1.0
G → A		−4.1 ± 1.1
G ← A		5.4 ± 0.9
average A → G	−4.0 ± 0.4	4.4 ± 0.9	8.5 ± 1.0	7.5 ± 2.4
G → S	−22.5 ± 0.5	−31.6 ± 1.9
G ← S	22.4 ± 0.5	37.4 ± 1.7
S → G		−35.7 ± 2.2
S ← G		−33.8 ± 2.0
average G → S	−22.4 ± 0.5	−34.6 ± 2.0	−12.2 ± 2.0	−8.4 ± 0.9
A → S	−26.2 ± 0.4	−30.2 ± 1.7
A ← S	26.3 ± 0.4	32.4 ± 2.0
S → A		33.5 ± 1.9
S ← A		−33.9 ± 1.4
average A → S	−26.3 ± 0.4	−32.5 ± 1.7	−6.3 ± 1.8	−0.9 ± 2.2
cycle	−0.2 ± 0.8	2.3 ± 2.8	2.5 ± 2.9	0

Individual estimates of ΔG_mutate were obtained by integration of the respective TI curves. Values for ΔΔG_bind were calculated from the differences between the averages over all available forward (black) and backward (red) transformations, as depicted in Figure S2. Experimental values are taken from reference [15], and error estimates were calculated from these data. Experimental cycle closure is 0 by definition. For clarity, only the mutated central residue of the tripeptide ligand is pointed out.