Symbolic Systems Biology is a growing area of research involving the application of formal logic-based methods to systems biology and bioinformatics. With biological data being acquired at ever increasing rates, symbolic approaches are being used more and more, in conjunction with numeric techniques, to help formalise expert knowledge and integrate information across different levels of biological abstraction. Recently, a number of symbolic approaches have been developed and usefully applied to a variety of biological problems. Such methods include

• formal logics (e.g., propositional/first-order/modal frameworks)
• computational logics (e.g., constraint/logic/answer-set programs)
• graphical models (e.g., Boolean/Bayesian/Petri nets)
• synthetic inference (abduction/induction)
• formal methods (e.g., model checking/pi-calculus/hybrid logic)
• qualitative reasoning
• action languages
• statistical relational learning

While many of these techniques are discussed in various texts on systems biology (see the useful links below), there has been very little inter-comparison these methods and their respective strengths and weaknesses. Instead we find a toolbox of isolated approaches with no conceptual foundations that allow any relationships to be theoretically explored and practically harnessed. Such a study is urgently needed to help to facilitate research in symbolic systems biology by providing a roadmap of which systems are best suited to which problems and allowing more effective exploitation and re-use of algorithms and data.

At the same time, there is also a need for more collaboration between numerical and symbolic biologists. This is necessary to better understand the advantages and drawbacks of the quantitative and qualitative approaches and progress towards a synergistic integration of the two. Ideally this should be done with advice from experimental biologists who are better informed of hot applications and emerging methods of data acquisition. This will help to provide a real context in which symbolic systems biology can be more usefully developed.

The primary aim is to initiate a systematic comparison of symbolic methods and their biological applications to better understand their strengths, weaknesses and supported features. For example, most methods operate upon networks which can be classified functionally (metabolic pathways, protein interactions, signal transduction, gene regulation, etc.) or formally (synchronous/asynchronous, discrete/continuous, deterministic/probabilistic, cyclic/acyclic, with/without feedbacks, etc.). We believe such features will help to construct a roadmap of what been achieved so far and suggest future directions for further research.

A secondary aim is to identify ways in which purely symbolic methods have been or could be combined with numerical techniques and applied to emerging problems in experimental and synthetic biology. We believe it is important to promote a dialogue between qualitative, quantitative, and experimental biologists. For example, it seems symbolic biologists can benefit from a better understanding of flux-based analyses used by numerical biologists; while numerical biologists can benefit from a better understanding of graph-based methods used by symbolic biologists. We believe both views are necessary to better handle real world noise and uncertainty.

Ideally, we will end up constructing a web site which provides an overview and comparison of the above-mentioned methods and applications. We would also hope to begin a discussion (possibly the topic of a future follow-on meeting) on how such methods could employ a standard data format (SBML/SGML) to allow a common specification language for such problems to facilitate inter-operability of the methods. We would also hope to produce some small and larger exemplars that could be used to compare the features offered by the various approaches and estimate their scalability to realistically sized problems. In this way we see the symposium as the first step towards providing a theoretical footing for the field of symbolic systems biology.

The symposium will be hosted by the National Institute of Informatics (NII) in Japan as part of its new series of Shonan meetings which provide a premier Asian location for informatics seminars following the successful European Dagstuhl format. These meetings aim to foster discussion of research and exchange of knowledge between world-class scientists, promising young researchers, and practitioners. They are held in the Shonan Village Center (near Tokyo) which offers a combination of facilities for conferences, training, and lodging in a resort-like setting (with a direct train connection from Narita airport).

We intend to encourage participation from a mix of symbolic biologists, numerical biologists, and experimental biologists. We will encourage these experts to explain their approach, demonstrate their tools, and participate in group discussions aimed at comparing the advantages and disadvantages of each approach and exploring the ways of integrating symbolic and numerical techniques.

A summary of the symposium with abstracts of the talks are available as a pdf

The following figures derived from the pre-workshop questionnaire (all of which are further explained in the above report) show (i) a clustering of participants based on the applications and methods with which they work and (ii) an association between those applications and the corresponding methods:

Coach to Kamakura (30 minutes) leaves at 13:30. Guided tour of Kotoku temple (Great Buddha), Hase temple, and Tsurugaoka Hachiman shrine. Coach to restaurant (Japanese style) arrives at 18:00. Banquet. Coach to Shonan (30 minutes) arrives at 20:30.

• Takehide Soh, Graduate University for Advanced Studies
• Junji Oshima, National Institute of Informatics
• Ayako Sumi, National Institute of Informatics

• Andrei Doncescu, CNRS
• Andrew Phillips, Microsoft Research
• Anne Poupon, CNRS
• Antonis Kakas, University of Cyprus
• Carolyn Talcott, SRI International
• Celine Rouveirol, University Paris 13
• Chitta Baral, Arizona State University
• Francois Fages, INRIA Paris-Rocquencourt
• Gauvain Bourgne, National Institute of Informatics
• Henning Christiansen, Roskilde University
• Hiroshi Hosobe, National Institute of Informatics
• Hiroshi Matsuno, Yamaguchi University
• Hisao Moriya, RCIS, Okayama University
• Katsuhisa Horimoto, Institute of Advanced Industrial Science and Technology
• Katsumi Inoue, National Institute of Informatics
• Koji Iwanuma, University of Yamanashi
• Marcus Tindall, University of Reading
• Mark Craven, University of Wisconsin
• Oliver Ray, University of Bristol
• Randy Goebel, University of Alberta
• Ross King, Aberystwyth University
• Sriram Iyengar, University of Texas
• Stephen Muggleton, Imperial College London
• Taisuke Sato, Tokyo Institute of Technology
• Takehide Soh, Graduate University for Advanced Studies
• Takeyuki Tamura, Kyoto University
• Tatsuya Akutsu, Kyoto University
• Tomoya Baba, National Institute of Genetics
• Torsten Schaub, university of Potsdam
• Toyoyuki Takada, National Institute of Genetics
• Wai-Ki Ching, University of Hong Kong
• Yoshitaka Yamamoto, Yamanashi University