Text S9

Domains 2 and 7, and assignment of enantiomer. Models were predicted for residues 26-87 and 357-404 using the ROSETTA algorithm [1-3]. The overall shapes of two of these models could be placed in the overall shapes in their corresponding electron densities (initially at 1.2σ contour) provided that the enantiomer of the phase set was flipped (phases were multiplied by -1). The vertical position of domain 7 was estimated from the number of dots in the dot models (Text S8). The volume of the dots was thought to be proportional to the number of amino acid residues (assuming that all 873 residues would be represented by density). In the initial partitioning shown in Fig. 1, domain 7 corresponds to density block 6. The initial models for domains 2 and 7 were later replaced with more credible models (Figs. 5b, 5e), but the shape-based enantiomer assignment was not changed.