All credit to Quasispecies for pointing this out, and I’ll happily concede authorship of this strategy.
Apparently, there is a correlation between successful lab candidates and a sharp melt curve- that is that, irregardless of how long the curve remains stable, the faster an RNA moves from its folded state to its denatured state, the better. So, the steeper the curve, and the less horizontal length the RNA takes to ‘melt’, the better, without regard to at what point the RNA begins to melt given that the RNA is stable under initial conditions.
If the horizontal length of the melt curve during denaturing is equal to 10 degrees Celsius, the score will be neutral. A point will be rewarded for denaturing in less than 10 degrees, a point will be deducted for each increment of 10 greater than a span of 10 degrees Celsius that the RNA takes to denature. For example, if the RNA ‘tops out’ or fully denatures over to course of 30 degrees Celsius, two points are detracted. If the RNA denatures over 7 degrees, it is issued a point. If the RNA denatures at 10-19 degrees, it is issued a neutral value.
Curious, now. I wasn’t aware of that. After looking it over briefly, it seems that we’d be able to take a harder look at it now. Personally, I expect that the sharper melt curves will do better than the ‘cap’ types- to a point, anyway- because they’ll denature quicker and reduce the ability of the RNA to form unintended bonds in the process, meaning that there are fewer acceptable forms of stability that the RNA can take on. I’m interested to see how it preforms as a strategy, and as well as in the lab- if my ‘hard boiled’ RNA or especially Quasispecies’ ‘1FMN hardboiled’ gets chosen, the point behind each being an intentionally abbreviated denaturing period.
I believe I’ll keep looking into this in the meantime. Thank you for the link.
Paramodic - The structure seems to dictate the shape of an “ideal” melt plots / specific heat curve. In “The Finger,” that pattern was a single sharp peak / abrupt melting transition. The finger was a single, long stack. In “The Branches,” practically all designs scoring above 95 have two high-temperature maxima, or a single peak with a “shoulder” in the specific heat curve. “The Branches” had two symmetric sections attached to a central multiloop.
It would be interesting to see if you could predict that “ideal” shape from the structure. You’re dealing with melting intervals, though, so maybe your strategy can avoid the problem of trying to find this shape.
Alternatively, you could look for Cp curve shapes / melting intervals that are known to be associated with bad designs. After all, the “ideal” shape in Cp curves is sometimes present in unsuccessful designs, but “bad” shapes are practically never present in successful designs. On top of that, “bad” shapes don’t seem to vary with the target structure. Wildly fluctuating baselines < 50 C, multiple maxima < 10 kcal / mol-K, and low melting transitions all seem to predict failure pretty reliably.
Data for “The Branches”: excel
You might have to actually view in excel, the formatting gets messed up in Google Docs. All designs have the Cp curve plotted individually. I’ve made combined graphs for the “winning” designs," and split them into categories. Very cluttered still. I need a better way of presenting all of that data. The other thread I posted on got a data update fairly recently, so you can look at that too.
Very interesting, I didn’t know he had looked into this already. The articles AlanRobot posted might make a few things clearer as well. Thanks for posting the link.
I’d still like to see this tested as a strategy, if possible. I’m excited to see how the soft-boil entry I’ve posted is going to turn out in the lab. Initial expectations are that it won’t do as well as a hard-boil design, but I can’t know for sure.