|FMN||Sarcin-ricin (15 nts)|
The top half is structurally quite similar, but the bottom differs a bit. Specifically, it is undeniable that the FMN molecule offers a much larger stacking area than the A+A platform in motif IL_85647.3
|FMN||Sarcin-ricin (13 nts)|
A possible way to compensate the difference would be to use a reduced form of the loop, with an additional cWW pair, like in motif IL_49493.4
By the way, this canonical pairs mimic...
I'm counting -11.9 kcal, much more than the -4.9 that we're supposed to mimic...
I understand that the mimic is only meant to give us an idea of whether a particular switch has a chance to work or not, but overshooting that much will probably give us a lot of false positives...
This is much less, ΔG -2.4 kcal, which possibly undershoots a bit, but at least we'd have solid reasons to be more confident that the switches passing that test will actually work with the real FMN experiment.
Another example, with a predicted ΔG -3.2 or -4.1 kcal, depending on the base pair that precedes.
Based on Brourd's constructs, a few tests we could run:
These are not perfect, but at least, the EteRNA model thinks these folds are even money...
The idea is to compare the stability of the unbound FMN binding site against various sarcin-ricin loops. If the model is correct, the 16-21 and 28-33 segments should display roughly the same SHAPE protection signal, something around 50%. If the hypothesis that the sarcin-loop is nearly as stable as a bound FMN site is correct, a strong difference should be visible between those 2 segments.
It seems easier to stabilize the design when using 5 nts long stems, instead of 6.