The hairpin you have given us does not follow the chirality of the large majority of native 5-residue hairpins, as described in the beta-beta rule of the ideal design paper. This may make it hard to come up with a peptide that will fold as intended; that loop will be trying to turn the other way.
It's true that some of the loop residues are in odd conformations, but this is what was observed in the crystal structure of the antibody-viral glycoprotein complex. Although strained conformations are best avoided in totally de novo designs, we're hopeful that, given that this is a conformation that has been observed experimentally, we can reproduce it in artificial designs. But that's a good observation!
(As an aside, we are considering variant designs in which we use unnatural amino acid residues with reversed chirality at some of these loop positions. Unfortunately, Foldit is not currently set up to allow us to give you a puzzle with mirror-image amino acids, but we might try that at some point...)
Late to the chase, but segment 257 has type "M" or "ligand", which throws off many recipes. Looks like a normal segment to me, is this the intention?
I have another version of the same solution that shows 257 as a normal segment. I think it's the parent of one of the type "M" ones, so maybe it has something to do with sharing. Time to check feedbacks.
Curious -- it was a normal segment in the puzzle setup, but I am seeing some funny business in some, but not all, of the PDB files for some of the results. I'm not sure what caused the problem, but if we can track it down, we will.