Modelling activities in molecular biology face the difficulty of prediction to link molecular knowledge with cell phenotypes. Even when the interaction graph between molecules is known, the deduction of the cellular dynamics from this graph remains a strong corner stone of the modelling activity, in particular one has to face the parameter identification problem. This article is devoted to convince the reader that computers can be used not only to simulate a model of the studied biological system but also to deduce the sets of parameter values that lead to a behaviour compatible with the biological knowledge (or hypotheses) about dynamics. This approach is based on formal logic. It is illustrated in the discrete modelling framework of genetic regulatory networks due to Ren