The understanding of the factors governing the enzyme catalysis is one of the most important goals of biochemistry and biophysics. In this context the contribution of theoretical research might be of high relevance. However, despite the huge amount of proposed approaches, the modelling of enzyme reactions still represents a very difficult task and a definitive and conclusive theoretical-computational strategy is still far from being available. In this study, after a presentation of the main difficulties associated to a coherent and possibly rigorous modelling of these processes, we present a computational theoretical method specifically designed for addressing complex molecular systems eventually applied to a benchmark reaction: the initial proton transfer in Triosephosphate Isomerase. This latter species, termed as the 'perfect enzyme' because of its exceptional ability as catalyst, provides an excellent test for evaluating the different environmental and intrinsic determinants at the basis of enzyme activity.