Is Computational Biology a Reliable Science?

Systems biology, neuroscience and other disciplines in biology rely increasingly on computer simulations to study the properties of models of biological systems, ranging from molecular processes via whole organs to the complete human physiome. This raises the issue of how reliable these computer simulations are as a scientific tool. In our own field, computational neuroscience, most colleagues agree that published results based on simulations cannot be replicated in many cases, often not even by their original authors. Scientists in other fields have come to similar conclusions: “The vast body of results being generated by current computational science practice suffer a large and growing credibility gap: it is impossible to verify most of the computational results shown in conferences and papers ” [1]. In at least one case, and innocent software error forced the retraction of several high-profile publications [2], and several authors have pointed out that many computational scientists lack even basic software engineering skills [3, 4]. In response to this reliability crisis, several groups have proposed rather technical solutions, namely that all scientist in a field should use the same, thoroughly tested software [1], and create systems for the seamless documentation of scientific workflow from the script specifying a simulation model to the final figure appearing in a journal, and should share the source code of their simulations [5]. In our view, this addresses only the replicability of simulations [6]. While important, reforms that addres