I see a new display in the lab, “pairing probability plots”. After a quick Googling, I still have a question (and the answers to most of my questions all seem to be “that’s an open area”, but here I go anyway) - how strong is the science behind the idea that this kind of probability plot is really interesting?
It seems to me, an armchair duffer at best, that it *would* still be an open question about whether it would be A Good Thing for the RNA molecule to spend a lot of time in a particular configuration. First of all, we’re never all that sure of exactly what the working configuration(s) of these biomolecules really is, are we? Second, it clearly has to get in and out of the proper configuration, so there will always be degrees of freedom, and who’s to say that more isn’t better than less? Or, it could be that some tiny bit of the molecule needs to conserve shape strictly, while the rest of it, not so much.
Just random thoughts on the matter, after seeing the new display pop up on my screen! Thanx.
You are right - we can’t say for sure that having high probability of target shape is necessarily a good thing in nature.
However, we think it can be a one way to guess how likely your RNA will stay in the target shape when synthesized. The more probable your RNA will stay in the target shape, the more probable it’ll be synthesized correctly.
Of course, this is only our guess - perhaps there is a better way to interpret the plot. We’ll leave that to our players to figure out!
so, to (attempt to) clarify, the first challenge is even to get the synthesis to happen correctly, getting the synthesized result to fold correctly is a *second* challenge - and if that’s correct, is it in fact ever verified at all?
JRStern, there is no “first” challenge (getting synthesis to happen). We can mostly get RNAs synthesized. It’s the “second” challenge we are trying to solve. I’m saying that clean dot probability plot “may” suggest that the RNA have a better chance of folding correctly when synthesized. But again, as I mentioned in my previous comment, this is only an assumption, and there is no definite answer to this.
Great question. I just wanted to add that the pairing probability is, while definitely approximate, also completely internally self-consistent within the algorithm that calculates the “natural mode” structure.
The algorithm that “natural mode” invokes considers a number of highly possible structures for a given sequence, and this output merely reflects any variance within this sample of “highly likely” shapes.
If the energy parameters were perfect, and if the algorithm could actually explicitly consider all possible conformations, then theoretically the base pairing probabilities would be perfect too. But since we know these assumptions are likely false, you can just think of this as getting a more detailed output from the eterna algorithm. Maybe it will be useful as a diagnostic for certain types of design flaws, or maybe it will just show more clearly ways the algorithm fails, noone can say for sure until they start testing their designs
You also bring up a great point, that perhaps getting all of the molecules in a test tube to stay in one, single precise shape all-of-the-time and at all-costs is not a feasible (or naturally occurring) strategy, which is one possible reason to avoid all GC designs with very negative free energies. It may be better to just get most of them folded most of the time to some average structure. The dot plots won’t really help you with that because such “squishiness” is not well handled by the algorithm, it will claim that all GC designs are awesome. . . so there’s probably some window of usefulness that the top designs will figure out. . .(note - speculation on my part)
“…you can just think of this as getting a more detailed output from the eterna algorith.” If we could get a larger plot and possibly a darker shade of grey in the plots it would have been great! The ultimate thing would be to have an expandable plot up alongside the requirements windows, with a real-time plot changing as the design changes
