I saw Brourd mention that the lab tests can only heat the RNA up to 50 degrees Celsius. In the Bistable lab the lowest melting point is 67, and that design is terrible, as are all the designs with the lowest melting points in that lab.
So what will happen to designs that have melting points higher than 50 C?
Heh, I stand corrected. It appears they do increase the temperature of the RNA to a much higher temperature (My apologies) However, melting point is only a metric based on free energy, lower free energy, usually means higher melting point. The free energy of a design will differ based on loop to base pair content, the base pairs used in stacks, and the model used to determine the free energy of the design.
Is the design with the lowest free energy and highest melting point the one that will score the best?
No, normally designs with the lowest free energy and higher melting points are made up entirely of G-C pairs. It is difficult both to synthesize and sequence designs with high GC %. The normal goal is to use around 60% G-C pairs. However, with the switch labs, that may not be the case, since the RNA needs to be able to switch states, and 60% G-C may be too much to facilitate a successful switch.
Is the design with the highest free energy and lowest melting point the one that will score the best?
No, this can indicate high A-U %, which can be unstable. However, designs with high A-U % may actually be better for switches, as they would facilitate the ability to move between energy states better.
Ultimately, the melting point isn’t a terrible metric to use, as it is a general measure of stability. However, like dotplots, they run into the problem of integrating energy bonuses into their generation. In fact, for many designs, energy “boosts” can mask potential instabilities. Whether or not these bonuses actually provide enough stability to correct these instabilities is not well known, and something to look into for future designs. Hope this helps!
I’m wondering about GC’s and the dot plots. I saw how delicate placing GC’s was in the Lines lab (#5). Of the 5 designs with the best dot plots they had GC’s gingerly placed spaced around in different ways, only one had a side-by-side GC pair. No one was a clear winner, all five dot plots were about the same despite how different they all were.
But then in the Bistable and Hair Trigger labs the dot plot seems to be GC hungry, rewarding the designs the with the most GC’s and pointing toward a single optimal design. Both labs have two unmoving stacks that are the same in both states and I suspect that there is much less penalty for locking those down with all GC’s. But despite what the dot plot says people are shying away from this in both labs even though they are all voting for and modifying Jieux’s leading GC rich designs.
In particular I noticed that the dot plot in Hair Trigger liked both unmoving stacks to be checkered with GC’s. At first this seemed perplexing to me, they should interfere with each other, but then I realized that one stack is a hairpin surrounded on either side by pairs of locked G’s. Any pair of C’s in the hairpin stack will temporarily mismatch with it’s 2G locked neighbors. So it zips up quicker when it is checkered. And it zips up long before it can interact with the much more distant strands of the unmoving 4 stack, so the 4 stack can afford to be checkered too, since that is the best pattern to prevent mismatches with any other GC’s in Jieux’s design.
So how accurate is the dot plot in the Bistable and Hair Trigger labs?