CPD12 Computational Protein Design problem in MiniZinc.

 D Allouche, S Traore, I Andre, S de Givry, G Katsirelos, S Barbe, and T Schiex
 Computational protein design as a cost function network optimization problem
 In Proc. of CP-12, Quebec City, Canada, 2012 

 http://genoweb.toulouse.inra.fr/~tschiex/CPD/CP2012-rebuilt/

 From an optimization perspective, the energy of a given protein, defined by
 a choice of one specific amino acid with an associated conformation
 (rotamer) for each residue can be computed as a sum of local
 terms representing interactions between at most two amino acid
 residues. The energy can then be written as:
 E = E_{c} + \sum_i E(i_r) +\sum_i \sum_{j>i} E(i_r,j_s)
 where E_{c} is a constant energy contribution capturing
 interactions between fixed parts of the model, E(i_r) is the self
 energy of rotamer r at position i capturing internal interactions or
 with fixed regions, and E(i_r,j_s) the pairwise interaction energy
 between rotamer r at position i and rotamer s at position j. 
 All terms are measured in kcal/mol and can be pre-computed and stored.

 The CPD12 folder contains only the 10 instances solved in (Allouche et al, CP 2012).
 All energy terms have been rounded to the nearest integer value
 after being multiplied by 100 (instead of 10**8 as in the CP 2012 paper)
 
 The energy terms in minizinc data files are directly rounded to the nearest integer value
 (without being multiplied) to avoid any integer overflows.