MEASURE | Def. Exact GC | Def. No GC | MFE Exact GC | MFE No GC | WT |
|---|
BP DIST TARGET | 0.06 ±0.03 | 0.09 ±0.04 | 0 ±0 | 0 ±0 | 0 ±0 |
ENERGY MFE | –0.53 ±0.11 | –0.85 ±0.12 | –0.48 ±0.12 | –0.79 ±0.14 | –0.38 ±0.11 |
ENERGY TARGET | –0.46 ±0.12 | –0.78 ±0.14 | –0.48 ±0.12 | –0.79 ±0.14 | –0.38 ±0.11 |
ENSEMBLE DEFECT | 0.12 ±0.04 | 0.14 ±0.06 | 0.05 ±0.02 | 0.05 ±0.02 | 0.08 ±0.05 |
EXP BP DIST | 0.07 ±0.03 | 0.1 ±0.04 | 0.03 ±0.01 | 0.03 ±0.01 | 0.05 ±0.03 |
PROP NAT CONTACT | 0.93 ±0.04 | 0.9 ±0.06 | 0.96 ±0.02 | 0.96 ±0.02 | 0.92 ±0.05 |
POS ENTROPY | 0.14 ±0.05 | 0.14 ±0.05 | 0.13 ±0.05 | 0.12 ±0.05 | 0.2 ±0.11 |
GC CONTENT | 42.88 ±9.14 | 82.07 ±3.5 | 42.9 ±9.15 | 80.92 ±4.09 | 42.88 ±9.14 |
LN DUAL PROB | –95.4 ±21.03 | –51.43 ±11.98 | –102.73 ±22.81 | –59.52 ±14.64 | –117.11 ±25.94 |
LN PROB | -10.81 ±4.05 | –10.95 ±4.68 | –1.38 ±0.55 | –0.96 ±0.45 | –2.02 ±0.97 |
MH STR DIV | 0.08 ±0.03 | 0.08 ±0.03 | 0.07 ±0.03 | 0.06 ±0.03 | 0.11 ±0.06 |
VIENNA STR DIV | 0.05 ±0.02 | 0.05 ±0.02 | 0.04 ±0.02 | 0.04 ±0.02 | 0.07 ±0.04 |
- For each of the 500 wild type (WT) pre-miRNA sequences, RNAdualPF sampled sequences, either having exactly the same GC-content as the WT sequence (‘Exact GC’) or with no control over GC-content (‘No GC’). The designation ‘MFE’ indicates that the sampled sequences were subsequently filtered to retain only those, whose minimum free energy structure is identical to the MFE structure of the corresponding WT pre-miRNA; otherwise, the designation ‘Def’ is used to indicate the default output of RNAdualPF, without the subsequent filtering step. For each WT pre-miRNA sequence, RNAdualPF generated 2000 sequences for the default case Def (no subsequent filtering), and 500 sequences for the non-default case MFE, such that sample MFE structure is identical to WT MFE structure. Various measures were used to compare the properties of RNAdualPF sampled sequences to those of wild type sequences: BP DIST TARGET: length-normalized average base pair distance d
BP
(s
0,s
∗) between the MFE structure s
0 of sequences sampled by RNAdualPF and the target structure s
∗. ENERGY MFE: length-normalized average free energy E(s
0) of MFE structure s
0. ENERGY TARGET: length-normalized average free energy E(s
∗) of target s
∗ for the respective sequences. ENSEMBLE DEFECT: length-normalized expected Hamming distance to target s
∗ [20]. EXP BP DIST: length-normalized expected base pair distance to target s
∗ [9]. PROP NAT CONTACT: expected proportion of base pairs of target s
∗ that occur in the MFE structure, i.e. \(\langle \frac {|s_{0} \cap s^{*}|}{|s^{*}|} \rangle \). POS ENTROPY: average positional entropy [22]. GC CONTENT: average proportion of positions occupied by G or C. LN DUAL PROB: average natural logarithm of the dual probability exp(−E(a,s)/RT)/Z∗(s) that sequence a adopts the structure s. LN PROB: average natural logarithm of the probability exp(−E(a,s)/RT)/Z(a) that sequence a adopts the structure s. MH STR DIV: length-normalized Morgan-Higgs structural diversity [23]. VIENNA STR DIV: length-normalized Vienna structural diversity, called ensemble diversity in [14]. Values of all measures for default sampled sequences having GC-content within 5 % of wild type GC-content (not shown) are essential identical to those of exact GC-content control
