Puzzle 1134: complex core bonus
| Case number: | 671071-2001133 |
| Topic: | Game: Other |
| Opened by: | gitwut |
| Status: | Open |
| Type: | Question |
| Opened on: | Thursday, September 3, 2015 - 22:23 |
| Last modified: | Monday, September 7, 2015 - 03:24 |
I've shared puzzle "gw7506.319 complex core bonus? why?" to scientists. The designed protein and it's clone are not close enough to warrant a complex core bonus, but yet it shows the condition as being met. If it's not a mistake, is the extra filter really necessary?
A per-residue bonus is enticing, and I agree it would be a little easier to interpret. I think, though, that this would be contrary to our aims.
We'd like for the puzzle filters to be applicable to a wide range of folds, and we have enough difficulty encouraging diversity already. I expect this idea—a graded reward for core residues—would drive players even more toward a single fold that maximizes core residues. And then we would miss out on a ton of perfectly good folds that happen to have fewer core residues.
You could make a strong point that adding core residues tends to increase protein stability. However, we're not interested in the most stable protein design; we're interested in all the protein designs that are stable enough to fold up.
Thanks for sharing, gitwut! These kinds of designs are really useful in helping us calibrate puzzle rules. The filter requirements on this puzzle are probably a little out of whack.
We're still trying to figure out the best way to encourage tight hydrophobic packing around H-bond networks. In previous puzzles, where the Core Exists: Complex filter was more stringent (see Puzzle 1074), it seems players had a really difficult time designing any H-bonds at the interface without violating the filter. We hope this setup at least gives you the freedom to optimize your H-bond networks.
I would also encourage you to experiment with different folds. Although your fold may satisfy the complex filter without making an interface, it may not make an optimal H-bond network. Someone else could find a fold that makes a much better H-bond network, but just barely passes the core filters.
Same problem with me, I wrote something to it as a comment on the puzzle page.
In general, what do you think about establishing a Core-exists-per-residue-filter? That means, the more residues are in the core, the more bonus you get. Wouldn't that be more reasonable (and less prone to errors) than the current filter setup?