See the comments for Puzzle 1446 for background on what's going on.
In puzzle 1503, the ligand at segment 321 is the designer molecule BIBP-3226. Similar to protein sidechains, it has multiple rotamers, different shapes or positions it can take.
In the selection interface you can select the ligand, the move through the rotamers using the left and right arrow keys.
The protein backbone in puzzle 1503 is all locked, but many of the protein sidechains can move. Since backbone is locked, remix and rebuild won't work, but shake and wiggle will.
Here's what the ligand looks like in a JMol (a molecule viewer):
BIBP-3226 seen in JMol: The ligand in puzzle 1503 is the designer molecule BIBP-3226.
Here's a similar pose seen in Foldit:
BIBP-3226 seen in Foldit.
You may notice that there are a few spots where JMol shows double bonds where Foldit has a single bond, and vice versa. The scientists have said that this is something to with how Foldit shows resonances, cases where there aren't quite enough electrons go around. Check out resonance (chemistry) on wikipedia for more.
Here's the version with atom numbers, as originally seen in the puzzle 1446 comments.
BIBP-3226 atom numbers.
Is there a scoring logic (scoring rules ?) behind the various different interactions that the ligand can meke with the rest? I wonder if some connections are favored over others based solely on which sidechain interacts with which atoms, thus being scored differently.
Clashing, voids and other factors will likely play a role too.