University of Surrey Institute of Advanced Studies

Microbial Systems Biology

Workshop Report

The aim of this workshop was to explore recent advances in Systems Biology, as applied to microbes. Microbes have been at the forefront of post-genomic research, largely for the simple reason that, having small genomes, their genome sequences have been easier to obtain than multicellular organisms. More than one hundred complete genome sequences for microbes are now available. The ready availability of detailed annotated genome sequence information along with high-throughput functional genomic data has stimulated novel approaches in microbial Systems Biology – that are currently unfeasible in more complex organisms.

The workshop was held on Thursday and Friday 14-15 July, 2005 at the University of Surrey. There were 35 participants, from Holland (1), Germany (2), Italy (1), UK (30). Twelve papers were presented, summarised below.

The theme for the first morning of the meeting was the construction and use of metabolic networks. The session was chaired by Johnjoe McFadden and opened with a talk by David Fell who described how metabolic networks could be used to predict metabolic capability of cells. He illustrated his talk with an analysis of pyruvate metabolism in E. coli and emphasised the need to constrain models with empirical data.

Steffen Klampt described the use of the FluxAnalyser tool and the application of flux balance analysis and minimal cut sets to predict the robustness of metabolic networks to both perturbation and mutation.

John Pinny described the use of a new network analysis tool “SHARKhunt” that can be used to automatically construct metabolic networks from genome sequence data.

The morning’s session concluded with a one hour’s discussion, which was a very lively session in which all participants shared ideas on the value and limitations of metabolic networks. Several points were highlighted, including the urgent need for a single agreed annotation of compounds and enzyme reactions.

The afternoon’s session concentrated on modelling of processes taking place in single cells and was chaired by Hans Westerhoff. The first talk, by Andrzej Kierzek described modelling single cell dynamics in E. coli. He argued that the low levels of regulatory proteins in each cell inevitably leads to stochastic fluctuations in the expression of proteins and thereby the cell’s phenotype. He demonstrated that these stochastic phenotypic fluctuations may be propagated from mother to daughter cell through non-genetic mechanisms.

Muktar Ullar described how MatLab could be used to perform simulations of regulatory dynamics.

The next talk was by Maria Schilstra who described the use of the NetBuilder graphical tool to model networks of transcriptional regulation and examine how stochastic fluctuations of levels of regulatory components can be propagated through the network.

The formal talks closed with another lively session that debated the importance of stochastic phenomena in living systems and particularly whether natural selection could have evolved inherently stochastic systems.

Discussions continued into the evening with an excellent workshop dinner that was attended by most participants.

The first session on Friday morning was chaired by Colin Smith and opened with a talk by Karen Lipkow who described the use of the Smooldyn Stochastic simulator to model chemotaxis dynamics in E. coli. She demonstrated how the rates of diffusion of regulatory proteins through the cell can profoundly influence the cell’s behaviour.

Duccio Cavalieri described the use of the Pathway Processor to model regulatory dynamics. He discussed how microarray transcriptional data may be integrated into these models. He also described the Reactome project which is a collaboration between the Cold Spring Harbor Laboratory, The European Bioinformatics Institute, and The Gene Ontology Consortium to develop a curated resource of biological pathways and reactions.

Lorenz Wernisch described the use Bayesian statistical methods to model transcriptional control in cells. He demonstrated how Bayesian regulatory networks can be “unrolled in time” to model time course data.

The session followed once again with a discussion session that focussed on how experimental data can be most efficiently integrated into transcriptional models.

The final session examined the application of systems biology techniques in real microbial systems and was chaired by David Fell. The afternoon opened with a talk by Claudio Avignone-Rossa which demonstrated the use of flux balance analysis to optimise antibiotic production in streptomyces.

Elaine Holmes described how the products of microbial metabolism may be detected by metabonomic analysis of clinical samples. The resulting data may be used to investigate the nature and dynamics of microbial communities in the patient’s gut and even to predict disease states.

Hans Westerhoff gave the final talk of the meeting which explored the full range of Systems Biology approaches applied to microbial systems. The emphasis of the talk was to go beyond static steady-state models to modelling real dynamic systems. These second generation models required knowledge of enzyme parameters, but Professor Westerhoff demonstrated that this approach was feasible for sub-systems of well-studied organisms, such as E. coli. The eventual aim of this approach was to produce a Silicon Cell that could be used, for instance, to predict potential drug targets.

The main themes that emerged from the meeting were:

  • The need for standard nomenclature in annotating enzyme reactions.
  • There was a strong feeling that to avoid the fate of previous theory-rich computational biology approaches (e.g. complexity research) it was imperative that Systems Biology models are continually interrogated with real biological data. 
  • New tools and methodologies were required to move from steady-state to dynamic systems.
  • New high-throughput biochemical approaches were needed to generate the kinetic data for dynamic models.
  • There was continual need to explore the importance of stochastic processes in living cells, but once again, it was important to constrain the models with experimental data.
  • New statistical approaches were needed to cope with the data emerging from transcriptional, proteomic and metabolic studies, and integrate these into systems biology models.
  • Modelling systems need to be made as user-friendly as possible to engage mostly computer illiterate biologists and to prevent a divergence of the discipline into parallel experimental and theoretical sciences.
  • There was a general feeling that Microbial Systems Biology was an exciting area of research.
  • There was enthusiasm for a second meeting next year. The favourite choice of venue was the beautiful city of Florence.

The organisers and participants are very grateful to:

  • The program committee:
    • Johnjoe McFadden
    • Graham Stewart
    • Claudio Avignone-Rossa
    • Mike Bushell
    • Colin Smith
    • Andrzej Avignone-Rossa
  • Madeline Brownett for her superb organisation skills.