I’ve come to the conclusion that of the three available energy models, Vienna2’s predictions fit the best with the riboswitch data we’re getting. If there is a general consensus on that, I suggest changing the default model from Vienna to Vienna2 for all lab puzzles.
From what I have seen so far, I agree with your observation, Omei.
Regularly NUPACK can’t show the switch, eg both states looks the same in native mode.
I also like the idea with making a default engine. It would be really nice if one could choose which one wanted as ones default.
Adding another configuration option is attractive, but the more options there are, the harder it is to find the ones that are really important. I wonder, is there anyone out there who would prefer to see the old Vienna as their default?
Although it would save me from the mistake of not changing the energy model each time I started a new design, I envision the primary beneficiary to be the newer player who hasn’t reached the level of sophistication yet to consult multiple energy models.
Agree. This will simplify to the new, which is main importance. And since Vienna2 for now seems the best, I would love this option.
I prefer NuPACK. It’s slower, sometimes crashes, but it does the trick, and there are certain predictions that it seems to be better at compared to the other engines and parameters.
In addition, it’s going to be the default for multi-strand puzzles, so may as well get players familiar with the NuPACK parameters.
Soon there will two categories of labs:
a) those that allow designers to select the engine
b) those that don’t (multistrands can currently only be computed by Nupack)
Because like Omei, I prefer the least amount of options whenever possible, I propose to implement a simple hidden persistence (stored as Flash cookie, so it would be per browser): when entering a lab of type (a), the automatically selected engine would be the last one the player used in that category.
Would this be satisfactory for everyone?
Beautiful solution, Nando.
That’s interesting, Brourd, that you prefer it. Do you have some examples handy that show why?
Persistence would certainly satisfy my needs, and I would think that would be true of most any experienced player. Would you also change the initial value to Vienna2, or wait for hard evidence that it is a better predictor in the lab?
Also, are the miRNA labs examples of multistrand labs, or is this something all new?
Great idea Omei and perfect solution Nando! I am with V2 but perhaps I need to look more into theNupack thing. I do hate how it slows things down, even just toggling thru the different models my computer just moves like molasses .