This thesis reports the analysis on the search algorithm PSICO (Processing Structural Information with Constraint programming and Optimisation) , an algorithm that uses constraint programming techniques for solving the protein structure determination problem from the set of pair distances collected by NMR. PSICO has shown great potential comparing to other solvers, but preliminary tests showed that its performance is highly dependent on good variable and value heuristics. This work was mainly motivated by the need to understand the effect that these and other important parameters have on search in this specific problem, aiming to improve the enumeration phase. Promising results were obtained concerning the final solution quality and algorithm robustness.
