Amazon currently lists well over 100 print-format books with ‘Systems Biology’ in the title, a four-fold increase in four years. Whatever the reasons for this publishing boom, would-be users of such books face a daunting task identifying something appropriate. Even ‘Computational Systems Biology’, the title phrase of Wiley's latest offering (Lodhi & Muggleton, Reference Lodhi and Muggleton2010), is rather uninformative, shared with a score of other books. It is, however, indicative that this collection of 17 contributions (corralled under rather broad section headings: ‘Overview’, ‘Biological Network Modelling’, ‘Biological Network Inference’, ‘Genomics and Computational Systems Biology’ and ‘Software Tools for Systems Biology’) comes from a computer-science rather than biological perspective – the editors are both in Imperial College, London's, Department of Computing. That is not to say that non-computer-scientists should steer clear – there is an element of the dating agency here – available techniques seeking partners in biological problems. I, for instance, was interested by the chapter on ‘membrane computing’, an area of biologically inspired computing new to me. At the same time, this book is not particularly biologist-friendly. Many of the chapters stand very close to their subjects and sometimes this is enough to turn off the general reader: The sub-section on inductive logic programming (ILP) starts by ‘review[ing] the notion and terminology used in ILP’, something potentially rather useful but, in practice, the first sentence of that explains one jargon term (‘a literal’) in terms of another (‘an atom’) and gets more involved from there. Not starting far enough back is a more surprising issue for the first, ‘overview’ chapter. While collating useful references, this misses the opportunity to present an overview of any of the various things that (computational) systems biology might mean, or indeed of the rest of the book. This could be an issue when many would-be readers' understanding of computational systems biology derives, directly or indirectly, from Kitano (e.g. Kitano, Reference Kitano2002), with his emphasis on biological simulation. The areas covered in this book are at once broader, for instance encompassing reviews of codon bias and biomedical imaging, and more specific, focused on what is often only the first step of systems biology, network construction.
Amid close-focus chapters, Jeremy Gunawardena's piece on ‘Models in Systems Biology’ is a welcome broad horizon. Addressing the ‘parameter problem and the meanings of robustness’ it brings perspective and clarity to a confused area of ‘traditional’ simulation-based computational systems biology. That there are problems with the multiplication of parameters is ably demonstrated by the following chapter where the 13 pages of main text are followed by 18 pages of appendix, primarily comprising the equations and parameters of the model presented. A machine-readable file in the biomodels database (Li et al., Reference Li, Donizelli, Rodriguez, Dharuri, Endler and Chelliah2010; http://www.biomodels.net/database/) would have been a whole lot more useful.
This is a diverse more than a comprehensive book, but its core is the section on ‘Biological Network Inference’, a theme that spreads out into other sections: Networks inferred via text mining appear in the ‘software’ section and two of the chapters in the ‘genomics’ section concern constructing transcriptional networks using genome sequences and expression data. The approaches are varied, even encompassing phylogenetic reconstruction. I'm not convinced that this latter chapter, which clearly presents the niceties of rate variation in biological sequence evolution, is particularly suited to a book on systems biology. However its presence makes the important point that phylogenetic trees are a form of biological network with a substantial history and body of theory, which has largely been ignored by, and ignored, systems biology up to now.
For those whose interest in systems biology doesn't involve constructing networks, perhaps taking them ‘off the shelf’ at KEGG (Kanehisa et al., Reference Kanehisa, Goto, Furumichi, Tanabe and Hirakawa2010), it is likely to be economically more sensible to download a chapter or two than buy this book. But even those uninterested in network construction can't avoid networks' pervasive presence in modern biology. So such a collection of network inference methods presents a challenge: do the questions asked of the resulting networks and the approaches used on them adequately account for the range of options and uncertainties that surrounds their construction, quite apart from real evolutionary variation in network structure (Knight & Pinney, Reference Knight and Pinney2009)? In simulation-based systems biology, uncertainty and change in network structure is rarely considered, the tools are not there to do so. I for one look forward to computational systems biologists providing such tools in the future. But those putting together books on the subject will need to find more informative titles if we're not to give up on sifting through the current morass of works labelled ‘systems biology’.
