Here is the sequence logo predicted by the SAM server:
H = helix
E = sheet
C = loop (or coil)
The taller the letter at each position, the higher the probability of that specific secondary structure for that amino acid.
For example, the Leucines at residues 71 & 76 are highly predicted to be part of a helix. However, the Leucines at residues 93 & 186 are not.
The above predictions are already on the starting model (based on the horizontal lines in the above figure).
Do you have some measure of how certain you are that the given portion is correctly matched?
We are pretty certain about the overall placement of the model. However, there is still a lot of uncertainty and side chain ambiguity in the modeled portion. Registry shifts might be one of them.
Given that residues 3-93 are identical to residues 6-186, is it safe to assume that the first half of the protein will fold the same and have the same contacts as the second half?
This is a designed protein and it basically was built in two different steps - The first one was building the helices, which indeed included symmetry!
The second step in designing this protein involved the design of loops to connect the helices, which might have broken perfect symmetry though.
So one would hope that residues 3-93 and 96-186 will fold up the same way... but you never know with proteins!