Many researchers at the interface of basic biological research and clinical application, and many clinicians too, will increasingly be faced with an unavoidable challenge of integrating systems biology approaches into their work. If the challenge is not met, we risk – among other things – producing a virtual human that is virtually incomprehensible in the clinical setting. Furthermore, we risk not integrating sufficient insights from the clinic into the model in the first place.
Biology has, until recently, been considered a ‘less difficult’ science than, say, physics or mathematics; but increasingly, leaders in the fields of systems biology realise that it is our inability to understand and model the true underlying complexity of whole biological systems (as opposed to their reduced parts) that is holding back deep physiological understanding of organisms: truly understanding biology involves getting to grips with unimaginable complexity, and some changes in paradigm along the way. And that is extremely difficult: it's not the same kind of challenge as faced particle physics, and that led to the building of the Large Hadron Collider, but it's more than a match for that problem in my opinion. Take systems that involve bidirectional causality, such as hierarchical biological systems. The realisation is sinking in that downward causality (organism-level phenomena influencing cellular and molecular phenomena) is as important to acknowledge and understand as upwards causality (from molecular phenomena up to organismal).
The diagnosis of a disease, for example, can occur at any level of the organisational hierarchy. Therefore it is seemingly important to know how that part of the hierarchy works – not in isolation – but in concert with the rest of the larger system in order to (1) identify the entire set of features (phenotype) of the disease and (2) be able to trace it downwards or upwards to the points in the larger system where a feasible treatment or cure might best be effected. Tracing upwards might, for example, lead to lifestyle or dietary modification as desirable and minimally invasive corrective strategies. Tracing downwards – e.g. to the identification of a genetic determinant – might, in future, indicate gene therapy. The value of an accurate physiological model of the human body lies in the ability to perturb it and understand what is ‘happening’. But that requires a substantial degree of familiarity with systems concepts – e.g. the ability to go into it at a variety of levels and correctly interpret the relationship between the variations from normal that occur in various regions of the system upon perturbation of one defined part.
On the one hand one might well ask ‘does the model really need to be that complicated?’ But if the rationale for personalised medicine is correct (i.e. everyone is slightly different in many ways that are important for disease outcome and treatment) then the answer is probably ‘yes’: the model needs to be as physiologically faithful as possible in order to be sensitive to the many individual variations that occur in the population at large. Basically we are talking about an in silico model that is almost as complicated as the in vivo organism. We will have progressed from the challenge of understanding the enormous complexity of a human being to that of understanding the enormous complexity of a computer model of a human being. At various stages of model development, specialists at discrete levels of the organismal hierarchy will develop the understanding of the levels above and below them necessary to integrate their models into the growing whole: that understanding needs then to be communicated more widely to build ‘community knowledge’ and engagement in developing and applying systems biology approaches closer to the clinic.