'The methodology of systems biology'

Douglas B. Kell and Hans V. Westerhoff

Science advances by an iteration between the world of mental constructs (ideas, background knowledge, hypotheses) and the world of sense data (experimental observations). Hypotheses on underlying mechanisms are induced from empirical findings. Consequences deduced by combining hypotheses with established underlying principles (such as the laws of chemistry and physics), are examined experimentally to test the new hypotheses. Given sufficient positive testing they are transformed to underlying principles through theorization. Much of biology has been concerned with the 'hypothetico-deductive' mode of reasoning that leads from ideas to data, without anchoring much into chemistry and physics. The far-from equilibrium nature, the dependence on nonlinearities and the complexity of many biological processes made the concepts of physics, which have been well elaborated and discovered for linear systems, less pertinent. The biology of entire living systems was considered by some to be too complex and ill-defined for hypotheses to be strict, testable and falsifiable.

Genomics has changed this picture drastically. Virtually all components that bring about living organisms are known, can be measured, and can be manipulated. Whether this reduces Biology to just another physical chemical science, with 'just' the same methodologies and quality criteria, is a fundamental issue. For, much of Life is associated with organizational and intelligence aspects that 'emerge' from molecular behaviour. Although these emergent properties are not in conflict with physics and chemistry, much of physics and chemistry traditionally shies away from complexity, hysteresis, and nonlinearity. Their paradigms favour the simplicity and Occam's razor that may not be relevant for Biology. We propose that this makes Systems Biology (the part of Biology that focuses on this issue) to be its own science with, indeed, its own methodology. We shall here contribute to the development of a philosophical basis for this new science by describing some of the means by which it operates in practice.

Systems biology consists of a principled interplay between experimental, computational and theoretical activities. It involves iterative relationships between the measurement of parameters and of variables. The recent developments in post-genomics have caused the 'Analytical' branch of Systems Biology (the 'omes') to develop most strongly, emphasizing measurement of the variables. We now need to redress the balance by transforming this empiricism into a hypothesis-generating arc that leads from data to knowledge. In the leaner, 'Synthetic' branch of Systems Biology, one typically starts with a qualitative ('structural') and often simple model of molecules interacting with each other in networks, then seeks to determine, and parametrise, the equations that describe these interactions. 'Bottom-up' methods start with purified entities (e.g. proteins) that allow the measurement of the parameters, while 'top-down' methods seek to infer their values via 'reverse engineering' . We give an overview of the types of experimental techniques that are being used to carry out such measurements.

Ordinary differential equations are the workhorse for 'forward' modelling. Much more computationally demanding are the methods proposed for solving the inverse problem ('system identification' – deriving parameters from measurements of time series of variables). Together these may be used to calculate Life – i.e. to produce a silicon cell that will display the main properties of the real cell. The implications are unprecedented for the sciences: if there is any place in the natural world where qualitatively new properties emerge, this is Life. Making Life calculable makes this emergence calculable, which may constitute a philosophical contradiction.