This page contains a few notes on playing techniques I use. No particular order; they may hopefully help automate protein folding.
One fruitful way to look at the game (not the only one) is to try and 'think like water'. After all, plain water is a much better player than you and me.
This is something I use regularly; it can yield good results throughout a puzzle.
When to use it: either after a move that didn't yield the result you wanted (a frequent occurence), or after causing a clash on purpose (e.g. moving a sidechain), and nudge-wiggling to optimize the result.
What to do: instead of using the "To Best" button to go back to your best configuration and trying something else, reposition sidechains one by one in the position they have in your best configuration. This will lower the score, which is normal - just nudge-wiggle the protein and you may move past your best score.
To see which sidechains have moved between your current and best configuration, simply alternate between current and best using Undo.
Advanced use: Did you perform a move that improved your score? Then try repositioning sidechains in the positions they assumed in the previous best configuration. Then nudge and wiggle away to try and further improve the score.
Think like water: An intuitive way to look at that: a shock with a water molecule displaces a sidechain. The structure moves to accomodate the new position, optimizing some areas in the process. The sidechain springs back to its former, more natural position, and the structure adapts again, keeping the optimizations it acquired earlier.
Note that this technique can easily be automated.