R80.New Frontier 3
The results for Round 80 took a long time to come out. And it seems to have been a very difficult batch. In many cases, there was a large amount of reactivity errors, and unfortunately, New Frontier 3 was no exception.
There are only a minority of designs that seem exempt of reactivity errors, or have relatively few. Observing those ones inspired me following thoughts.
A confusing presentation
Obviously, there was a lot of confusion associated with the way I had chosen to present the challenge.
For instance, this design would probably have scored better if the pseudoknotting sequences had matched properly. So close...
In this case, I'm not sure whether it was confusion or deliberate choice, but the designer was consistently shooting for a slightly different target (which did work, apparently)
Another case of mismatched sequences.
Mismatched again. The result is possibly interesting though, for a rather remarkable SHAPE signature, reminiscent of the Semicircle 2 bends lab.
Not mismatched, but I'm confused by the choice of poly(U)...
Finally, the whole point of this confusing setup, was to force designers to create a pseudoknot in order to achieve success (score-wise), but the "fake" secondary structure I created for that goal was actually not impossible to solve directly (congratulations to JR), contrary to what I had expected.
One interesting case
This design had a subtle "flaw". The designer used a slightly different target, with the 5' strand of the pseudoknot shifted downstream by one nt, and the 3' strand shifted upstream also by 1 nt. The knotting sequences were a (rather) good match, so it could have worked just as well as the original target. But it didn't work too well.
I don't think that the presence of a UU mismatch in the pseudoknot could have presented a serious problem, but who knows? Maybe the 5' promoter (locked GG) interferred. Or possibly, the second shift caused the sequence to appear in a less convenient 3D position for an easy docking. And of course, all these factors may have played a role here.
It seems that it will soon be technically possible to create puzzles and labs with variable length segments. I think it will be specially interesting to use this new functionality in the context of pseudoknots.
A correct sequence was still not necessarily the promise of a winner though...