Figure 3.

MFE defect. For each target τ^k ∈ Pseudo for k = 1…201, the corresponding vertical bar represents the frequency (out of 30 trials) where MFE(ϕ^l, τ^k) = τ^k was reached. Comparison of performance of Enzymer (A) with the performance of MODENA (B) and antaRNA (C) shows Enzymer outperformed the other two methods in 191 and 194 cases respectively. A binomial test statistic with 99% confidence suggests Enzymer outperforms both methods in generating sequences with lower MFE defect. Furthermore, MODENA outperforms antaRNA in 127 cases and the binomial test statistic suggests superior performance of MODENA compared with antaRNA. Structure IDs on the x-axis are sorted based on increasing size of the corresponding targets.