We present a tool for the verification of qualitative biological models. These models formalise observed behaviours and interrelations of molecular and cellular mechanisms. During its development a model is continuously verified. Predicted behaviours are compared with behaviours observed in experim ental data. Moreover, the model must not exhibit behaviours which contradict existing knowledge about capabilities of the biological system under investigation. Model development is an iterative process involving many rounds of prediction, verification and refinement. Due to the complexity of biological systems this process is laborious and error prone, which motivates the development of “model debugging” tools. The qualitative models we investigate represent large-scale molecular interaction networks describing gene regulation, signalling and whole-cell metabolism. We integrate a steady state model of whole-cell metabolism with a dynamic model of gene regulation and signalling represented as a Petri net. This Quasi-Steady State Petri Net (QSSPN) representation allows the generation of dynamic sequences of molecular events satisfying substrate, activator, inhibitor and metabolic flux requirements at every state transition. The reachability graph of the dynamic part of the model is examined and for every transition in this graph the satisfaction of metabolic flux requirements is verified by well-established linear programming techniques. Our approach is based on network connectivity alone and does not require any kinetic parameters. We demonstrate the applicability of our method by analysing a large-scale model of a nuclear receptor network regulating bile acid homeostasis in human hepatocyte. To date, simulation and verification of QSSPN models have been performed exclusively by Monte Carlo simulation. Random walks through the state space were used to find examples of behaviour satisfying properties of interest. Here, we provide for the first time for QSSPN models an exhaustive analysis of the state space up to a finite depth, which is possible due to several effective optimisations. Contrary to the Monte Carlo approach, we can prove that certain behaviour cannot be realised by the model within a given number of steps. This allows rejection of models which are not capable to reproduce experimentally observed behaviours, as well as verification that biologically unrealistic behaviours cannot occur in the simulation. We show an example of how these features improve identification of problems in large scale network models.

We present a tool for simulation of populations of living cells interacting in spatial structures. Each cell is modelled with the Quasi-Steady Petri Net that integrates dynamic regulatory network expressed with a Petri Net (PN) and Genome Scale Metabolic Networks (GSMN) where linear programming is used to explore the steady-state metabolic flux distributions in the whole-cell model. Similar simulations have already been conducted for single cells, but we present an architecture to simulate populations of millions of interacting cells organized in spatial structures which can be used to model tumour growth or formation of tuberculosis lesions. For that we use the Spark framework and organize the computation in an agent based “think like a vertex” fashion as in Pregel like systems. In the cluster we introduce a special kind of per node caching to speed up computation of the steady-state metabolic flux. Our tool can be used to provide a mechanistic link between genotype and behaviour of multicellular system. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

MOTIVATION: Dynamic simulation of genome-scale molecular interaction networks will enable the mechanistic prediction of genotype-phenotype relationships. Despite advances in quantitative biology, full parameterization of whole-cell models is not yet possible. Simulation methods capable of using available qualitative data are required to develop dynamic whole-cell models through an iterative process of modelling and experimental validation. RESULTS: We formulate quasi-steady state Petri nets (QSSPN), a novel method integrating Petri nets and constraint-based analysis to predict the feasibility of qualitative dynamic behaviours in qualitative models of gene regulation, signalling and whole-cell metabolism. We present the first dynamic simulations including regulatory mechanisms and a genome-scale metabolic network in human cell, using bile acid homeostasis in human hepatocytes as a case study. QSSPN simulations reproduce experimentally determined qualitative dynamic behaviours and permit mechanistic analysis of genotype-phenotype relationships. AVAILABILITY AND IMPLEMENTATION: The model and simulation software implemented in C++ are available in supplementary material and at http://sysbio3.fhms.surrey.ac.uk/qsspn/.

Phenotypic differences of genetically identical cells under the same environmental conditions have been attributed to the inherent stochasticity of biochemical processes. Various mechanisms have been suggested, including the existence of alternative steady states in regulatory networks that are reached by means of stochastic fluctuations, long transient excursions from a stable state to an unstable excited state, and the switching on and off of a reaction network according to the availability of a constituent chemical species. Here we analyse a detailed stochastic kinetic model of two-component system signalling in bacteria, and show that alternative phenotypes emerge in the absence of these features. We perform a bifurcation analysis of deterministic reaction rate equations derived from the model, and find that they cannot reproduce the whole range of qualitative responses to external signals demonstrated by direct stochastic simulations. In particular, the mixed mode, where stochastic switching and a graded response are seen simultaneously, is absent. However, probabilistic and equation-free analyses of the stochastic model that calculate stationary states for the mean of an ensemble of stochastic trajectories reveal that slow transcription of either response regulator or histidine kinase leads to the coexistence of an approximate basal solution and a graded response that combine to produce the mixed mode, thus establishing its essential stochastic nature. The same techniques also show that stochasticity results in the observation of an all-or-none bistable response over a much wider range of external signals than would be expected on deterministic grounds. Thus we demonstrate the application of numerical equation-free methods to a detailed biochemical reaction network model, and show that it can provide new insight into the role of stochasticity in the emergence of phenotypic diversity.

UNLABELLED: Constraint-based modeling of genome-scale metabolic networks has been successfully used in numerous applications such as prediction of gene essentiality and metabolic engineering. We present SurreyFBA, which provides constraint-based simulations and network map visualization in a free, stand-alone software. In addition to basic simulation protocols, the tool also implements the analysis of minimal substrate and product sets, which is useful for metabolic engineering and prediction of nutritional requirements in complex in vivo environments, but not available in other commonly used programs. The SurreyFBA is based on a command line interface to the GLPK solver distributed as binary and source code for the three major operating systems. The command line tool, implemented in C++, is easily executed within scripting languages used in the bioinformatics community and provides efficient implementation of tasks requiring iterative calls to the linear programming solver. SurreyFBA includes JyMet, a graphics user interface allowing spreadsheet-based model presentation, visualization of numerical results on metabolic networks represented in the Petri net convention, as well as in charts and plots. AVAILABILITY: SurreyFBA is distributed under GNU GPL license and available from http://sysbio3.fhms.surrey.ac.uk/SurreyFBA.zip.

MOTIVATION: Taverna workbench is an environment for construction, visualization and execution of bioinformatic workflows that integrates specialized tools available on the Internet. It already supports major bioinformatics services and is constantly gaining popularity. However, its user interface requires considerable effort to learn, and sometimes requires programming or scripting experience from its users. We have integrated Taverna with OpenOffice Calc, making the functions of the scientific workflow system available in the spreadsheet. In CalcTav, one can define workflows using the spreadsheet interface and analyze the results using the spreadsheet toolset. RESULTS: Technically, CalcTav is a plugin for OpenOffice Calc, which provides the functionality of Taverna available in the form of spreadsheet functions. Even basic familiarity with spreadsheets already suffices to define and use spreadsheet workflows with Taverna services. The data processed by the Taverna components is automatically transferred to and from spreadsheet cells, so all the visualization and data analysis tools of OpenOffice Calc are available to the workflow creator within one, consistent user interface. AVAILABILITY: CalcTav is available under GPLv2 from http://code.google.com/p/calctav/ CONTACT: [email protected].

BACKGROUND: Constraint-based approaches facilitate the prediction of cellular metabolic capabilities, based, in turn on predictions of the repertoire of enzymes encoded in the genome. Recently, genome annotations have been used to reconstruct genome scale metabolic reaction networks for numerous species, including Homo sapiens, which allow simulations that provide valuable insights into topics, including predictions of gene essentiality of pathogens, interpretation of genetic polymorphism in metabolic disease syndromes and suggestions for novel approaches to microbial metabolic engineering. These constraint-based simulations are being integrated with the functional genomics portals, an activity that requires efficient implementation of the constraint-based simulations in the web-based environment. RESULTS: Here, we present Acorn, an open source (GNU GPL) grid computing system for constraint-based simulations of genome scale metabolic reaction networks within an interactive web environment. The grid-based architecture allows efficient execution of computationally intensive, iterative protocols such as Flux Variability Analysis, which can be readily scaled up as the numbers of models (and users) increase. The web interface uses AJAX, which facilitates efficient model browsing and other search functions, and intuitive implementation of appropriate simulation conditions. Research groups can install Acorn locally and create user accounts. Users can also import models in the familiar SBML format and link reaction formulas to major functional genomics portals of choice. Selected models and simulation results can be shared between different users and made publically available. Users can construct pathway map layouts and import them into the server using a desktop editor integrated within the system. Pathway maps are then used to visualise numerical results within the web environment. To illustrate these features we have deployed Acorn and created a web server allowing constraint based simulations of the genome scale metabolic reaction networks of E. coli, S. cerevisiae and M. tuberculosis. CONCLUSIONS: Acorn is a free software package, which can be installed by research groups to create a web based environment for computer simulations of genome scale metabolic reaction networks. It facilitates shared access to models and creation of publicly available constraint based modelling resources.

A general paucity of knowledge about the metabolic state of Mycobacterium tuberculosis within the host environment is a major factor impeding development of novel drugs against tuberculosis. Current experimental methods do not allow direct determination of the global metabolic state of a bacterial pathogen in vivo, but the transcriptional activity of all encoded genes has been investigated in numerous microarray studies. We describe a novel algorithm, Differential Producibility Analysis (DPA) that uses a metabolic network to extract metabolic signals from transcriptome data. The method utilizes Flux Balance Analysis (FBA) to identify the set of genes that affect the ability to produce each metabolite in the network. Subsequently, Rank Product Analysis is used to identify those metabolites predicted to be most affected by a transcriptional signal. We first apply DPA to investigate the metabolic response of E. coli to both anaerobic growth and inactivation of the FNR global regulator. DPA successfully extracts metabolic signals that correspond to experimental data and provides novel metabolic insights. We next apply DPA to investigate the metabolic response of M. tuberculosis to the macrophage environment, human sputum and a range of in vitro environmental perturbations. The analysis revealed a previously unrecognized feature of the response of M. tuberculosis to the macrophage environment: a down-regulation of genes influencing metabolites in central metabolism and concomitant up-regulation of genes that influence synthesis of cell wall components and virulence factors. DPA suggests that a significant feature of the response of the tubercle bacillus to the intracellular environment is a channeling of resources towards remodeling of its cell envelope, possibly in preparation for attack by host defenses. DPA may be used to unravel the mechanisms of virulence and persistence of M. tuberculosis and other pathogens and may have general application for extracting metabolic signals from other "-omics" data.

UNLABELLED: Taverna workbench is an environment for construction, visualization and execution of bioinformatic workflows that integrate specialized tools available through the internet. It is gaining popularity fast, because of supporting the most important bioinformatic services and its simple, yet robust graphical notation. Here we present XQTav-an extension of Taverna that provides full integration with XQuery (the query language for XML) engine. XQTav allows execution of XQuery scripts in Taverna workflow diagrams. All existing Taverna processors can be accessed in the XQuery scripts. This provides an alternative way of specifying subworkflows in Taverna and is useful when one deals with query-like algorithms (e.g. filters and inner joins). Moreover, XQtav may be used to automatically generate an XQuery script that is equivalent to Taverna's workflow. This constitutes another way of creating and enacting bioinformatic workflows: overall structure of a diagram is drawn in Taverna environment, XQuery code is generated and possibly adjusted by hand. It can be executed by XQuery engines or incorporated into other software environments. AVAILABILITY: XQtav is an open source software. It may be downloaded from http://xqtav.sourceforge.net/. The page also contains various tutorials and examples, including the one described in this report.

MOTIVATION: The availability of a huge amount of molecular data concerning various biochemical reactions provoked numerous attempts to study the dynamics of cellular processes by means of kinetic models and computer simulations. Biochemical processes frequently involve small numbers of molecules (e.g. a few molecules of a transcriptional regulator binding to one 'molecule' of a DNA regulatory region). Such reactions are subject to significant stochastic fluctuations. Monte Carlo methods must be employed to study the functional consequences of the fluctuations and simulate processes that cannot be modelled by continuous fluxes of matter. This provides the motivation to develop software dedicated to Monte Carlo simulations of cellular processes with the rigorously proven Gillespie algorithm. RESULTS: STOCKS, software for the stochastic kinetic simulation of biochemical processes is presented. The program uses a rigorously derived Gillespie algorithm that has been shown to be applicable to the study of prokaryotic gene expression. Features dedicated to the study of cellular processes are implemented, such as the possibility to study a process in the range of several cell generations with the application of a simple cell division model. Taking expression of Escherichia coli beta-galactosidase as an example, it is shown that the program is able to simulate systems composed of reactions varying in several orders of magnitude by means of reaction rates and the numbers of molecules involved. AVAILABILITY: The software is available at ftp://ibbrain.ibb.waw.pl/stocksand http://www.ibb.waw.pl/stocks. SUPPLEMENTARY INFORMATION: Parameters of the model of prokaryotic gene expression are available in example files of software distribution.