The biggest problem I'm seeing is that helices
strands and loops which dont have a template structure to align to
(since the similairty is 60% and the identity is 30%) are just floating
in space, they are not in a "natural" position. How do I fix that first?
If you don't have a template, Modeller isn't going to know where to put
them. It will do its best to enforce proper stereochemistry, but yes, I
would expect them to be floating in space. You would need additional
input information, such as distance restraints, to pin them down.
Secondly maybe it even is accepting the secondary structure restraint,
but it is severly beding the helices, since it wants to aggresivly align
to the template structure...with which parameters would I handle this so
I could get a reasonable structure?
Modeller will always make your model follow the template - that's how it
works. If you don't want to align part of the structure to the template,
alter the alignment so that those residues are aligned to gaps instead.