Due to the inherent limitations of wet-lab techniques, the experimental data regarding cellular signaling pathways often consider single pathways or a small subset of them. We propose a methodology for composing signaling pathways data in a coherent framework. Our method consists in specifying the signaling pathway as a computationally executable model. We rely on the timed concurrent constraint language ntcc to represent the system in hand as a set of stoichiometric-like equations resembling the essential features of molecular interactions. The main advantages of our approach stem from the use of constraints (formulas in logic) and from modeling of discrete time clocks in ntcc. We can deal with partial information, representing the fact that several features of the biological system may be undetermined. We can explicitly represent the time needed for a reaction to occur. We model and simulate some well known cross-talking networks, such as the TNFα, the EGF and the insulin signaling pathways as well as their interactions.