Abstract: The modelisation of large biological systems such that metabolic networks or gene interaction networks, implies interpretations of heterogeneous biological knownledge. A common feature of observations accumulated by biologist on these systems is their qualitative nature. More over it often happens that the knownledge is based on comparisons between different experimental conditions. In previous papers variations in observed variables between two conditions was interpreted in terms of equilibria shift. This leads to use qualitative equations in a sign algebra. The present report gives details on the implementation and main algorithms involved.