Building complex multimeric models

[Francois Sampieri <sampieri.f ( ( at ) ) jean-roche.univ-mrs.fr>]

 Hi all,
I had to construct a model of a protein complex derived from
 experimental 3D structures. I believed that it could be obtained
 without special skills.
 The templates were:
1- A protein P made up of two well separated domains covalently
 attached: A and B, such:
                   A-B
 2- A non-covalent complex between 3 protein subunits C, D, and E:
                 C==D==E
My purpose was to split the protein P into its two domains and to
 covalently attach each domain to the C and E units with a polypetidyl
 arm:
           A       and       B
          /                   \
         C                     E
 to eventually obtain, in the presence of D, a new complex:
                  A=B
                 /   \
                C==D==E
 with A=B bound non-covalently.
I worked with InsightII-Biopolymer from MSI, and I succeded in
 splitting the starting protein model into its two separate domains,
 then building the arms (I'm so proud of that  ;-), but a problem
 arose when I attempted to bind these arms to their respective C/E
 unit, since when creating the covalent bond between, say the A-arm
 and C, the whole C==D==E complex went into a very embarrassing
 orientation in the space, due to the rigid geometry of the bond
 created. Seems as if there is only one way to make a new bond in
 InsightII.
 I tried lots of methods without success and now I wonder if this
 could be possible with this program.
Unless I missed some hidden tricks in InsightII, my question is: is
 there any other program (running on a SGI O2; IRIX 6.5) capable of
 building bonds with 'soft' (not rigid) geometry?
 Thus, it would de possible to build the complex with the actual
 conformations and orientations of A-B and of C==D==E bound to the
 polypeptide arms, then energy-minimize the arms to make them to
 conform to the standard geometry.
 A program with only building capabilities would be satisfactory,
 provided that it should be able to create pdb files and handle more
 than 15.000 atoms.
 I hope I'have not been too obscure.
 Thanks and best wishes to all!
 Francois
 Francois Sampieri
 sampieri.f ( ( at ) ) jean-roche.univ-mrs.fr
 Laboratoire de Biochimie
 IFR Jean Roche
 Faculte de Medecine Nord
 MARSEILLE
 FRANCE