Briefings in Bioinformatics Advance Access originally published online on July 18, 2007
Briefings in Bioinformatics 2007 8(4):208-209; doi:10.1093/bib/bbm036
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Integrative biology — the way forward
Let me begin by thanking the editors for the opportunity to be the guest editor of this themed issue on Integrative Biology. It is to their credit that they have recognised the importance of these approaches and have been willing to set aside both the December 2006 and this issue for what are closely related, emerging areas. Indeed, the distinction between Integrative Biology and Systems Biology (December's theme) is extremely subtle. The latter concerns the development of computational and mathematical models of the structure and behaviour of biological systems with a view to reaching a quantitative understanding of them. The former has a greater emphasis on the process of developing such models — integration is what we aspire to, but acknowledge that we are not there yet.Integrative Biology advances our understanding of biological phenomena through the close collaborative efforts of laboratory and theoretical scientists who develop mechanistic mathematical models that lead to hypotheses for laboratory testing. The cycling between theoretical and laboratory work leads to models that have predictive value. It differs from Systems Biology in that there may be integration in various dimensions: laboratory with theoretical scientists, models that span different scales (such as molecular up to cell scales or beyond), or models that integrate different types of physics or mathematics. Perhaps this subtle difference between means and ends is what prompted the UK's BBSRC to announce calls for Centres for Integrative Systems Biology.
Industry has always been in a position to enforce such multidisciplinary working, and many times over this has led to novel solutions to problems that have substantial Intellectual Property and hence commercial value. It is no surprise, therefore, that the major academic research funding agencies are pushing universities hard to break down traditional departmental boundaries and to create organisational structures that bring together people from radically different backgrounds. The institutions that succeed in doing this are likely to do better in the current funding climate, where collaborative research has clearly become the order of the day. This is excellent news for bioinformaticians who are well used to operating in such an environment. The only minor shock to us might be the even greater breadth of people to deal with. When scoping a new project to develop models that integrate molecular up to field and farm scales, I recently found myself coming into contact with social scientists and practitioners of arguably the first ‘omics’ area, namely econ-omics!
This new climate opens up many new areas and challenges for Bioinformatics. Earlier work outlines initial thoughts in this regard [1]. In short, there are new types of static and time-series data, including images from microscopes, body scans and closed-circuit cameras; electrophysiology and mechanical properties of biological materials. These generate even larger datasets than the current ‘omics’ techniques, and will require novel approaches to managing and manipulating them. GRID technologies and new uses of high-performance computing clusters are very likely to play a role here.
The integration of these different scales will also require new semantic integration. Ontologies are unlikely to be entirely satisfactory, as the different communities are unlikely to abandon their longstanding terminologies. While Description Logic looks like a good solution to this issue, it will require much effort to establish comprehensive connections between the various domains of knowledge. Arguably, the biggest unmet challenge in this area is the development of a general framework for multiscale-model development. At present, we have bespoke methods specific to individual models.
I believe bioinformaticians also have an educational and possibly even ambassadorial role to play in this arena. We already have the advantage of some familiarity with the languages of biology, maths and computing, and of the mindsets and motivations of the people in these disciplines. We can, therefore, act as interpreters and motivators, outlining the priorities of biologists to the modellers, while helping biologists to abstract their systems of interest into forms that are amenable to modelling approaches.
There can be a stepwise approach to introducing biologists to integrative systems approaches. For the most part, their conceptual models are usually networks readily represented as graphs in the mathematical sense. From here, they can pick up the concept of dynamics from discrete techniques such as Petri nets or P-systems. It is then a comparatively short conceptual step to the continuous approaches of differential calculus. While most mathematicians would far rather dive straight in with the latter, as they perceive it to be the most representative of the system, I suspect biologists would prefer to tread the path outlined above, especially as it is likely to fit in with the course of their laboratory work: post-genomic data leads to networks, time-series experiments show the dynamic changes which eventually need to be described quantitatively.
This issue of Briefings in Bioinformatics should be viewed very much as a follow-on from and expansion of the December issue and Merilli et al. [2]. The order of the manuscripts has been arranged so that the manuscripts refer to approaches on progressively higher and across progressively broader physical scales. Chaouiya et al. outlines the ways in which Petri nets have been used to construct and simulate the dynamic behaviour of biological networks, while Thieffry describes dynamic regulatory circuits (which represent the first level of physiological regulation). Mjolsness outlines the use of fugacity for modelling the partitioning of macromolecules between different multimolecular complexes, while Burrage et al. describe techniques for modelling processes that take place in biological membranes. These involve among other things diffusion and stochastic events.
Scale integration is more pronounced in the next two papers. Thorne et al. show how agent-based modelling can be used for tissue patterning. These integrate molecules able to diffuse comparatively quickly between the ‘cells’ (agents) which then respond in various ways that lead to patterns at the ‘tissue level’ (the set of agents as a whole). Poirier and Iglesias describe perhaps the most spectacular array of integrative modelling techniques, as they study mechanosensation. The final manuscript provides a case study of how computational and laboratory scientists have worked together to improve predictive methods for and the discovery of transcription factor binding sites.
This issue also includes a review of the Handbook of Bioinspired Algorithms and Applications. For many years, computer scientists have marvelled at the ways in which biological systems can estimate things and come to optimal solutions to engineering problems. It is little wonder, therefore, that they have developed software that attempts to emulate brain behaviour, evolution, social behaviour etc., and this review provides a fuller background. I hope you find this issue informative and are encouraged both to look further at this growing area of endeavour and to be an ambassador to (or at least a mediator between) people in the different disciplines involved.
Centre for Plant Integrative Biology
Multidisciplinary Centre for Integrative Biology
School of Biosciences, University of Nottingham, UK
References
- Hodgman TC, Ugartechea-Chirino Y, Tansley G, Dryden I. The implications for bioinformatics of integration across physical scales. J Integrative Bioinformatics (2006) 3:39.
- Merilli E, Armano G, Cannata N, et al. Agents in bioinformatics, computational and systems biology. Briefings in Bioinformatics (2007) 8:45–59.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||