Fig. 3

a The distribution of Lysine riboswitch secondary structures projected to the Hamming distance from the reference structure. Both the probabilities and their logarithm are plotted. b The secondary structure representations of alternative structures. c The alternative structures of the the Hamming distance range [9,56] (left), [56,82] (middle) and [82,95] (right, “alt3”). Terminator and antiterminator sequences are marked with circles (see the main text.) Figures were generated by RNAplot in the ViennaRNA package [7]. d The 2D distribution of Lysine riboswitch secondary structures projected to the Hamming distances from the structure predicted by CentroidFold and the alt3 structures, calculated by using RintD [12]. The probabilities were converted to the free energy (i.e. −RT logP) and were plotted. (left) The case with the wild-type RNA sequence. (right) The case with the M1 mutant. Secondary structures that have low Hamming distances to the alt3 structure are more stable with the M1 mutant than those with the wild-type (marked with an arrow). Free energies greater than 20 kcal/mol are plotted in white. Note the checkered pattern appears due to difficulties in achieving some Hamming distance constraints