Functional Genomics and Systems Biology.

Different people define these research approaches very differently (or sometimes avoid defining them at all). This is partly fair enough, because this approach calls upon an unprecedentedly broad range of skills, models, techniques, and approaches. So, with the caveat that my concept / definition may be quite different from those of others, who are just as entitled to their own definitions, I will briefly describe what the systems biology approach is all about (to me).

What systems biology isn't:

The philosophy of this approach is not to select a single, or a particularly limited subset of, cellular components to study, based upon existing models and pre-conceptions of cellular function in order to perform mechanistic studies. Nor is it to dissect in detail the molecular mechanisms that are key or particular to a cellular function of interest. At the same time it does not deny or exclude such specific approaches, when they are identified by the system-level investigative strategy. In this sense, systems biology is not a ‘re-badging’ of molecular biology, and nor is it limited to mathematical / theoretical modelling of living systems, genome sequencing, bioinformatics, transcriptional profiling, protein structural biology, systematic or focussed mutagenesis, or any other single methodological approach, although it can reasonably be expected to call upon any or all of these activities, in various combinations, in the process.

What systems biology is:

The ‘systems’ approach, as I conceive it, involves the study and interrogation of complete function systems, as an integrated whole using a variety of strategies to determine those cellular components and behaviours that are critical to the processes in which we are interested (which are detailed specifically below). Suitable systems can be particular pathways or responses, the whole cell, comparative studies of different cells, or even comparison and analysis of the wider population and availably gene-pool. Indeed, an integrated systems biology approach will include such a ‘vertical’ stratification of concepts and approaches, as well as a minimally exclusive ‘breadth’ of data collection and testing at each level. This conceptual approach is philosophically similar to the concepts that underlie methods such as ‘signature tagged mutagenesis’ (STM), and in ‘in-vivo expression technology’ (IVET), in that the study of how the organism functions is not constrained by our assumptions and misunderstandings, and we allow the study of behaviour to inform our decisions on which genes or cellular processes to focus upon.

How it tends to work:

Systems biology is frequently an iterative process. For example, expression profiling in response to a condition of interest will identify one or more candidates that are important in controlling the response and / or in directly mediating the adaptation to the new environment. These genes become candidates for knock-out and structural / functional analysis. The way in which the mutants behave is investigated, including the use of further expression profiling, which both adds to our understanding of the actions of that regulator or effector protein (hopefully), and identifies further genes for targeted investigation. It is also an approach that calls for overlapping strategies to investigate the system. For example, comparisons of behaviourally different species and strains at a sequence and gene-complement level leads to the identification of genes that can be prioritized when they are identified in expression studies. Likewise, genes that have features of phase variation, suggesting that they have phenotypes associated with particular selective niche conditions, can also be preferentially targeted, when they are found to be different between behaviourally distinct populations, or part of potentially host-adapted responses. Similarly, genes that have been horizontally acquired may be the focus of investigations of differing species and strain behaviours. Comparisons of the structures of proteins that are associated with different behaviours or regulatory responses might also be prioritized on the basis of strain or species differences, or their presence in different global responses, or specific regulons identified in knock-out experiments. All of these things, and others, combine to contribute to a systems-level investigation of an organism or process.

You can't be in a hurry, but the only way should be forwards.

From the above description, one thing should be evident, and it is frequently not appreciated by those who have not participated in such projects. In one aspect this approach can generate a great deal of information. However, any one investigation frequently identifies many genes / proteins of interest. The (wise) systems biologist will be reluctant to triage the list of genes that are considered potentially important within a system, but will also realize that it is not possible (or probably useful) to pursue all of the candidates identified in any single study. So, the results from different experiments and approaches are progressively combined, compared, and new experiments developed that progressively move towards more specifically informative experiments. Thus the approach, despite its high ‘information content’, is not one that can be (productively) rushed – and it is frequently a slower path than many expect. However, perhaps above all others it has one compelling advantage over the highly reductionist approaches built upon very specific models and hypotheses. If a useless vaccine candidate or drug target is intensively pursued and then abandoned, or a physiological or biochemical model is found to be completely wrong, then while this adds some insight into what not to use or what models to abandon, it does not necessarily take the investigator nearer their goals. Well designed ‘systems biology’ experiments should normally be progressively informative, and even when they show a cherished model to be wrong, they will generally provide insights into how things actually are. In truth, the most common thing that functional genomics and systems biology experiments reveal is that our current models are frequently very partial or completely wrong, and that overly focussed reductionist approaches have successfully built up some concepts that are deeply flawed.

A hypothesis?

Perhaps controversially I would contend that a systems biologist does not absolutely need a hypothesis (although it’s most certainly not contra-indicated), but s/he does need a goal and / or question, and a sufficient understanding of the system and tools to be able to construct and perform informative experiments.

Systems biologists vs. reductionists?

Are systems biologists ‘anti-reductionists’? I would most emphatically say "No." to this question. There should be no conflict between the ‘reductionists’ and the ‘systems biologists’. Systems biologists are using a non-reductionist approach to the observation and quantitation of biological processes, with a view to gaining an understanding of how a system works that is no less oriented to its fundamental and mechanistic components than those with a ‘traditional’ reductionist approach. If people considering themselves to be systems biologists become investigators who no longer generate specific and testable hypotheses (and cease to have this as a purpose) then they are no longer truly scientists at all (I refer anyone who doesn't concur to the works of Karl Popper). The ultimate goal of systems biology must be the generation and testing of the new models that the non-reductionist approach enables them to construct – although there is nothing wrong with building models that others can and are perhaps better equipped to test. Modelling and data collection that does not lead to testable, and ultimately tested, hypotheses is broadly a waste of time and resources. Without complementary reductionism, systems biology would be pretty pointless. What systems biology is supposed to achieve is better models, and a better information context in which to test them.