I apologize for being a bit naive, but it turns out that I'm working an low resolution mode (3.6 A). A great part of the chains have previously been built with polyalanine, since there's obviously no density for the side chains. When using phenix, I obtain a model just as good with all residues mutated to the real ones (with a high resolution reference model). Other low resolution structures in my field have been refined with no alanine mutations. Is the polyalanine mutation actually still in use?
