| | 100% over 25 nt | 100% over 30 nt | ≥95% over 40 nt | ≥94% over 50 nt | ≥88% over 60 nt |
|---|
<100% over 25 nt
| - | N = 0 P = n/a | | | |
<100% over 30 nt
| N = 21 P = 0.06 | - | N = 25 P = 0.37 | | |
<95% over 40 nt
| | N = 1 P = n/a | - | N = 1 P = n/a | |
<94% over 50 nt
| | | N = 13 P = 0.32 | - | N = 42 P = 0.01 |
<88% over 60 nt
| | | | N = 0 P = n/a | |
- N is the number of pairs in a given cell and P the likelihood probability of being subject to off-target cross-reaction. We chose not to compute the likelihood probability when N was below 10 pairs. If the likelihood probability is high (i.e. in the vicinity of 0.3), the fractions of WT/WT, WT/PH and PH/PH for the selected pairs do fit our model (see text) and are therefore likely to be subject to off-target cross-reaction. From this table, we predict that efficient RNAi is observed for pairs sharing high percentage identity over more than 25 nt and less than 60 nt. High identity on 40 nt (without high identity on 30 nt or 50 nt) fits well with efficient RNAi, however since the data used here are limited, we think it is reasonable to propose that RNAi is maximized when sequence similarity is very high in the range 30–50 nt. We are aware that the counts we used here can be sometimes low; in the future, the use of a larger RNAi data set should allow to gain in statistical power.
