Table 2.
Total calculation time in seconds (predicting secondary structures for 17 γ values of 850 RNA sequences) of the estimator appro3-1 for each δ and each combination of the probability distributions
| δ | pa-ct | pa-mc | pc-ct | pc-mc |
|---|---|---|---|---|
| 0 | 276 869 | 290 579 | 274 758 | 290 174 |
| 0.0001 | 16 471 | 5107 | 16 393 | 5009 |
| 0.001 | 7962 | 3202 | 7897 | 3124 |
| 0.01 | 3836 | 2130 | 3781 | 2065 |
| 0.1 | 2552 | 1600 | 2502 | 1541 |
We set α = 1/(|D| + 1) = 0.1 in this experiment. ‘x-y’ means appro3-1 with x of p(a) and y of p(s). pa, pc, mc and ct denote the ProbAlign model (Roshan and Livesay, 2006), the ProbCons model (Do et al., 2005), the McCaskill model (McCaskill, 1990) and the CONTRAfold model (Do et al., 2006a), respectively.