Is there a reason segments 1-25 aren't frozen?
Yes. We wanted Aβ to be able to wiggle with the design.
Unlike larger, highly structured proteins (e.g. the ebolavirus glycoprotein), the Aβ peptide is very flexible, and its structure will be highly influenced by its binding partner. We expect the Aβ peptide to wiggle a bit when it binds, in order to better fit the binding pocket of the designed protein.
is our goal to make this bind in such a way that the AB does not want to bond with cholesterol or lipids? that it is unable to bind with anything else once in the pocket?
The goal is to create a tight-enough binder to the Aβ peptide such that it will not interact with other molecules, and to keep it from aggregating with other Aβ molecules.
Wouldnt it be better if the AB segments would not be included in the filter?
Generally speaking, yes, it would probably be best to keep the target protein in its starting conformation, since we know that it is already capable of binding in this arrangement. Unfortunately, the Fragment Filter code needs some updates before we can implement this feature, but those are forthcoming!
As I've mentioned above, though, the Aβ peptide is very flexible in solution, and is capable of taking many different conformations. Our goal is to create a stable complex between Aβ and the designed protein, which may entail finding a new binding conformation for Aβ. In this line of thinking, we probably want the entire complex to look like an "ideal" protein, in which all components pass the Fragment Filter.
Perhaps a foolish question, but in a normal circumstance what does the binding look like...i.e. where on the AB molecule is the binding occurring, is it just the sheets or is it also the loop connecting the sheets?
I'm assuming you're asking about how Aβ binds to itself in the disease state? One of the problems that makes Aβ such a difficult molecule to study is that we don't know exactly how it self-associates!
We do know that Aβ can aggregate as a "cross-beta fibril," in which many Aβ molecules take a β-hairpin conformation and form an extensible β-sheet. However, we don't know how it associates to form smaller, soluble complexes which are thought to be most toxic to neurons.