The production and implementation of methodologies allowing the construction of coherent, precise and trustworthy predictive biological models has become an inescapable necessity. The financial stakes -attached to this reality are very high indeed, be it in the public or the private domains (health, research, pharma and foodstuffs industries, environment, etc.). Modelling biological systems is widely presented as a problem in computational sciences. While certainly very true for low complexity, practically continuous systems, this view cannot be upheld in the case of discontinuous, hyper-complex systems such as living entities. In these domains, modelling becomes a problem in biology assisted by computational sciences and certainly not the obverse. The following article will attempt to demonstrate why it is so, using concrete examples.