Table 2.

Calculated changes in binding energies of DNA and amino acid point mutations compared with experiments (15,30)

DNA Sequence	ΔE binding	ΔG binding (experiment)
Binding to wild-type PurR
seqc	0	0.0
seq3	0.16	0.8
seq1	2.02	1.6
seq2	6.78	3.2
Binding to the K55A mutant
seq1	−7.53	−0.06
seq2	−4.47	−0.46
seqc	0	0
seq3	1.13	0.5
Mutant	ΔE binding	ΔG binding (experiment)
PurR mutants bound to the consensus sequence
WT	0	0
L54M	5.79	0.38
L54S	16	larger, not measured
L54T	10.05	,,
L54V	6.15	,,
K55A	12.55	3.48

When only the DNA is mutated, the binding order is correct (top panel). When both DNA and the protein are mutated (middle panel), two DNA mutants are lower in binding energy and one higher than the original sequence. This is correctly identified by our method, but the binding preference to seq₁ and seq₂ is reversed. When only the protein is mutated, the binding preferences of the DNA to the mutants are correctly captured (bottom panel). Energies are given in kilo calories per mole and measured relative to the respective consensus protein–DNA complex.