Poster No. 1: Stability and timing in Boolean dynamics

Fakhteh Ghanbarnejad and Konstantin Klemm
Bioinformatics, University of Leipzig, Leipzig, Germany

Regulatory dynamics has mathematical descriptions in terms of rate equations for continuous variables and, after discretization of the state space, as Boolean maps. Here we study this discretization in detail. In particular, we define the stability of a Boolean state sequence in consistency with the stability of the original continuous trajectory that has been discretized. In essence, the stability criterion translates infinitesimal perturbations in the state space of the continuous system into infinitesimal time lags in the Boolean counterpart. For a class of randomly connected systems with randomly drawn Boolean functions, so-called Kauffman networks, we find that the dynamics is stable for almost all choices of parameter values. The so-called "chaotic" regime in Kauffman networks appears only as a damage spreading effect after flip perturbations. We conclude that regulatory systems amenable to state discretization do not exhibit chaotic behaviour.

Poster No. 2: Boolean versus continuous dynamics on simple two-gene modules

Eva Gehrmann
Technische Universitat Darmstadt, Hochschulstr. 664289 Darmstadt Germany

We investigate the dynamical behavior of simple modules composed of two genes with two or three regulating connections. Continuous dynamics for mRNA and protein concentrations is compared to a Boolean model for gene activity. Using a generalized method, we study within a single framework different continuous models and different types of regulatory functions, and establish conditions under which the system can display stable oscillations. These conditions depend only on general features such as the ratio of the relevant time scales, the degree of cooperativity of the regulating interactions, and the logical structure of the interactions. Our results combine and generalize the findings of several disconnected previous studies.

Poster No. 3: Sparseness in model gene regulatory networks

Marcin Zagorski
Marian Smoluchowski Institute of Physics and Mark Kac Complex Systems Research Centre, Jagellonian University, Reymonta 4, 30-059 Krakow, Poland

Gene regulatory networks typically have low in-degrees, whereby any given gene is regulated by few of the genes in the network. They also tend to have broad distributions for the out-degree. What mechanisms might be responsible for these degree distributions? Starting with an accepted framework of the binding of transcription factors to DNA, we consider a simple model of gene regulatory dynamics. There, we show that selection for a target expression pattern leads to the emergence of minimum connectivities compatible with the selective constraint. As a consequence, these gene networks have low in-degree, and ``functionality'' is parsimonious, i.e. is concentrated on a sparse number of interactions as measured for instance by their essentiality. Furthermore, we find that mutations of the transcription factors drive the networks to have broad out-degrees. Finally, these classes of models are evolvable, i.e. significantly different genotypes can emerge gradually under mutation-selection balance.

Poster No. 4: Receptor cross-talk in angiogenesis: Mapping environmental cues to cell phenotype in a stochastic, Boolean signaling network model

Thimo Rohlf (1,2) and Amy L. Bauer (3)
(1) Epigenomics Project, Genopole, Genopole Campus 1 - Genavenir 6, 5 rue Henri Desbruères, F-91030 ÉVRY cedex, France>
(2) Max-Planck-Institute for Mathematics in the Sciences, Inselstr. 22, D-04103 Leipzig, Germany
(3) Theoretical Division, Los Alamos National Laboratory, Los Alamos, 87544, USA

Cancer invasion and metastasis depend on tumor-induced angiogenesis. Based on experimental data, we propose a Boolean signal transduction network model highlighting the cross-talk between key receptors involved in angiogenesis. We identify relationships between receptor activation combinations and cellular function, and show that cross-talk is crucial to phenotype determination. The network converges to a unique set of output states that correspond to known cell phenotypes: migratory, proliferating, quiescent, apoptotic, and it predicts one phenotype that challenges the “go or grow” hypothesis. Finally, we use the model to study protein inhibition and to suggest molecular targets for anti-angiogenic therapies.

This study [1] is the first to propose a signal transduction network model that couples VEGF-RTK, ITG, and cadherin receptor signaling cascades, to capture receptor cross-talk during angiogenesis. We use this model to investigate how cellular behavior depends on and is controlled by changing environmental signals. Both for discrete and continous time state space, Boolean network dynamics converges to a unique set of output states that correspond to known cell phenotypes: migratory, proliferating, quiescent, apoptotic, and it predicts one phenotype that challenges the “go or grow” hypothesis. Signal transduction is very robust against molecular noise, with one exception: a mixed feedback scheme between RhoA and Rac1, as sometimes discussed in the literature, is extremely sensitive to noise, leading to erreatic cell movement. After a “shock” (randomization of cell-internal concentrations), we find a transient apoptotic response that increases with molecular noise, suggesting a dynamical mechanism that ensures that when the system is under stress or shocked, that regeneration occurs preferentially from the most healthy cells, while unhealthy cells are eliminated. Finally, we use the model to study protein inhibition and show that, in principle, it is possible to suggest molecular targets for anti-angiogenic therapies.

References:

1. 1. Bauer, AL, Jackson, TL, Jiang, Y. and Rohlf, T., J. Theor. Biol., 264, 838-846 (2010)

Poster No. 5: Gene regulation in time and space: interdependence between genome structure and gene network dynamics

Thimo Rohlf (1,2), Ivan Junier (1,3) and François Képès (1)
(1) Epigenomics Project, Genopole Campus 1 - Genavenir 6, 5 rue Henri Desbruères, F-91030 ÉVRY cedex, France
(2) Max-Planck-Institute for Mathematics in the Sciences, Inselstr. 22, D-04103 Leipzig, Germany
(3) Institut des Systemes Complexes Paris Île-de-France, 57/59, rue Lhomond 75005 Paris, France

Gene regulatory networks (GRN) at transcription level represent a crucial step of cellular information processing. While data suggest a highly structured spatial organization of co-regulated genes, most GRN models completely neglect spatial effects. Here, we investigate the interdependence between genome structure, GRN topology and GRN dynamics in the framework of an artificial genome model. Our study suggests that local concentration effects, for example induced by positional regularities of co-regulated genes, considerably reduce noise and increase robustness of regulatory dynamics. Further, we investigate evolution of positional regularities under different types of mutation operators.

Data on the spatial organization of genomes, obtained both from experiments and bioinformatics, suggest strong positional regularities of co-regulated genes on chromosomes. For example, it has been shown that the genes which are regulated by a given transcription factor (TF) and the gene coding for this TF tend to be located periodically along the DNA [1]. It was suggested that a solenoidal organization of chromosome structure [2] can explain the periodic pattern by distance minimization in 3d space. Using a thermodynamic framework based on a polymer model of DNA with sparse, interacting transcription sites, it was further shown that periodic gene positions, for different topologies of chromosome structuring, can favor the formation of transcription factories [3]. These regularities suggest that cells exploit local concentration effects induced by spatial proximity of co-regulated genes to optimize the efficiency of gene regulation [4,5,6].

Here, we investigate the effects of spatial genome organization on the topology and dynamics of GRN within the framework of an artificial genome model [7]. Our study suggests that local concentration effects can considerably reduce noise and increase robustness of regulatory dynamics. Using evolutionary algorithms, we also show how evolution may exploit these effects for cellular information processing.

References:

1. 1. Kèpès, F. (2004), J. Mol. Biol. 340, 957-964
2. Képès, Francois and Vaillant, C. (2003), Complexus 1 (3), 171-1803.
3. Junier, I., Martin, O. and Képès, F. (2010), PLoS Comp. Biol. 6, e1000678
4. J. M.. G. Vilar and S. Leibler (2003), J. Mol. Biol. 331, 981–989
5. Lanctot, C., Cheutin, T., Cremer, M., Cavalli, G. and Cremer, T. (2007), Nat. Rev. Genet. 8, 104-115
6. Manceny, M. , Aiguier M., Le Gall, P., Junier, I., Hérisson, J. and Képès, F. (2009), BICoB 5462, 270-281
7. Rohlf, T . and Winkler, C. (2009), Adv. Comp. Sys. 12, 293-310

Poster No. 6: n.n.

Poster No. 7: Evolutionary design of oscillatory genetic networks

Yasuaki Kobayashi, Tatsuo Shibata, Yoshiki Kuramoto, Alexander S. Mikhailov
Fritz-Haber-Institut der Max-Planck-Gesellschaft (Berlin), Germany

The present study is devoted to the design and statistical investigations of dynamical gene expression networks. In our model problem, we aim to design genetic networks which would exhibit stable periodic oscillations with a prescribed temporal period. While no rational solution of this problem is available, we show that it can be effectively solved by running a computer evolution of the network models. In this process, structural rewiring mutations are applied to the networks with inhibitory interactions between genes and the evolving networks are selected depending on whether, after a mutation, they closer approach the targeted dynamics. We show that, by using this method, networks with required oscillation periods, varying by up to three orders of magnitude, can be constructed by changing the architecture of regulatory connections between the genes. Statistical properties of designed networks, including motif distributions and Laplacian spectra, are considered. Also, we show that the same evolutionary optimization method can be used to increase the robustness of the networks against several types of noises.

References:

1. Y. Kobayashi, T. Shibata, Y. Kuramoto and A. S. Mikhailov, European Physical Journal B, 76, 167-178 (2010)

Poster No. 8: Models of biological network evolution

Richard Stein & Hervé Isambert
Institut Curie, Physico-Chimie Curie, UMR 168 (CNRS-UPMC), 11 P. & M. Curie, 75005 Paris, France

About forty years ago, Susumu Ohno proposed gene and genome duplication as a key source of novel protein functions in the course of evolution [1]. Using the example of protein-protein interaction networks, Evlampiev & Isambert [2] proposed an analytical approach to study the topology and conservation of a non-oriented/homogeneous network under a genomic duplication-divergence process. We have found that this non-oriented/homogeneous network model can in fact be extended to model the duplication-divergence evolution of oriented/heterogeneous networks, such as signal transduction pathways and regulatory networks. The evolutionary phase diagram of such oriented/heterogeneous network model can be solved analytically in the asymptotic limit of large growing networks. It can also be applied to analyze, in particular, the different retention biases of different gene functional classes following either segmental duplication or whole genome duplication events. We also investigated the effect of biological-relevant network constraints such as the finite genome size constraint for prokaryotes [3]. With the above steps, we try to improve our understanding of the role of gene duplication in the evolution of biological networks, leading to predictions on the conservation of network motifs and experimentally verifiable features of biological networks.

References:

1. Ohno, S. (1970): Evolution by gene duplication, Springer.
2. Evlampiev K., Isambert H. (2008): Conservation and topology of protein interaction networks under duplication-divergence evolution, Proc. Natl. Acad. Sci. U. S. A., 105, 9863–9868.
3. Isambert H., Stein R. R. (2009): On the need for widespread horizontal gene transfers under genome size constraint, Biology Direct, 4 (28).

Poster No. 9: Long-term Evolutionary Constraints on Signaling Pathways Implicated in Cancer Development

Param Priya Singh, Hervé Isambert
Institut Curie, Physico-Chimie Curie, UMR 168 (CNRS-UPMC), 11 P. & M. Curie, 75005 Paris, France

Gene duplication and subsequent mutations are believed to play a major role in evolution. Duplication creates raw genetic material on which positive, negative as well as neutral selections (i.e. natural and purifying selections as well as genetic drift) operate to create novel functions and/or sub-functions. This continuous process of duplication-divergence transitions was, in particular, central in the emergence of metazoans, vertebrates, and mammals from relatively simple unicellular organisms [1]. Among the most striking events in the evolutionary history of these organisms are whole genome duplication (WGD) events, which have now been firmly established in all major eukaryote kingdoms. WGDs are rare but dramatic genome doubling transitions leading to a transient two-fold genetic redundancy that is eventually greatly reduced, as only about 10-20% of the genes retain a WGD duplicate (coined “ohnolog”) 300 to 500 million years (MY) after WGD. In particular, comparative genomic studies have now demonstrated that the common ancestor of all vertebrates has undergone two rounds of WGD about 500 MY ago. These WGDs as well as smaller segmental duplications have shaped vertebrate genomes and biomolecular networks. They bring not only genetic novelties but also evolutionary constraints that restrict by construction the emerging properties of biomolecular networks. The effects of such evolutionary constraints are either independent from the functions of individual genes (i.e. by merely stemming from the duplication divergence process itself [2–5]) or may lead to different expansions of gene families depending on their actual functions.

In the present study, we investigate such a function-dependent evolutionary constraint. Namely, we explore the evidence supporting the role of deleterious oncogenic mutations on the long-term evolution of signaling pathway networks implicated in cancer development and progression. The consequences of oncogenic mutations leading to cancer progression and metastasis are typically studied on time scales of months up to years, i.e. within the life span of an organism. Yet, because of their adverse consequences, oncogenic mutations might have also biased the expansion and emergent properties of signaling pathways implicated in cancer on long evolutionary time scales, i.e. on hundreds of millions of years. We have analyzed, in particular, the consequences of deleterious oncogenic mutations on the evolutionary fates of gene duplicates arising from WGDs. Our preliminary results suggest that signaling pathways implicated in cancer have been largely shaped by the two WGDs near the time of divergence of vertebrates with significant retention of ohnologs. Furthermore, oncogenes in these pathways exhibiting functional auto-inhibitory constraints appear to have retained more ohnologs than the ones which do not have such constraints suggesting that expansion and emergent properties of signaling pathways implicated in cancer have largely resulted owing to such non-adaptive auto-inhibition constraints rather than adaptive (beneficial) innovation.

References:

1. Ohno S: Evolution by Gene Duplication. New York: Springer 1970.
2. Evlampiev K, Isambert H: Modeling protein network evolution under genome duplication and domain shuffling. BMC Systems Biology 2007, 1:49.
3. Evlampiev K, Isambert H: Conservation and topology of protein interaction networks under duplication-divergence evolution. Proceedings of the National Academy of Sciences 2008, 105(29):9863–9868.
4. Cosentino Lagomarsino M, Jona P, Bassetti B, Isambert H: Hierarchy and feedback in the evolution of the Escherichia coli transcription network. Proceedings of the National Academy of Sciences 2007, 104(13):5516–5520.
5. Isambert H, Stein R: On the need for widespread horizontal gene transfers under genome size constraint. Biology Direct 2009, 4:28.

Poster No. 10: Evolution of the AMP-activated protein kinase controlled gene regulatory network

Karin Breunig, Ivo Grosse
Institute of Biology, Martin-Luther-University Halle-Wittenberg, Germany
Institute of Computer Science, Martin-Luther-University Halle-Wittenberg, Germany

Alterations in transcriptional regulation are considered major driving forces in divergent evolution. This is reflected in different species by the variable architecture of regulatory networks controlling highly conserved metabolic switches. The switch from glycolysis to gluconeogenesis and back that serves as an adaptive mechanism in response to changes in nutrient or oxygen supply in all living systems is controlled by a conserved set of protein kinases and their down-stream effectors. Apparently, the wiring of these regulators has changed gradually during evolution. Here, we present a project within the priority program Information and Communication Theory in Molecular Biology that aims at uncovering sequential steps in this evolutionary process and at deepening our understanding of the evolution of the transcriptional regulatory network controlling gluconeogenesis.

Poster No. 11: De-novo motif discovery using parsimonious Markov models

Ralf Eggeling, Andre Gohr, Jens Keilwagen, Pierre-Yves Bourguignon, Michaela Mohr, Stefan Posch, Andrew D. Smith, Ivo Grosse
Institute of Computer Science, Martin-Luther-University Halle-Wittenberg, Germany
Leibniz Institute of Plant Biochemistry (IPB), Halle, Germany
Leibniz Institute of Plant Genetics and Crop Plant Research (IPK), Gatersleben, Germany
Max-Planck Institute for Mathematics in the Sciences, Leipzig, Germany
Molecular and Computational Biology, University of Southern California, USA

Deciphering transcriptional regulatory networks is one of the challenges of systems biology, and the identification of functional DNA binding sites of transcription factors, enhancers, or insulators is a necessary prerequisite. Existing algorithms for the identification of DNA binding sites are often based on the oversimplified position weight matrix model, which makes the biologically questionable assumption that all nucleotides within a binding site are statistically independent.

Here, we propose a de-novo motif finding algorithm that relaxes this assumption by using parsimonious Markov models and a DP algorithm for maximizing the a posteriori probability. These models generalize Markov models, and the DP algorithm is capable of inferring the optimal set of dependencies within binding sites based on the observed data. Hence, they capture strong statistical dependencies while neglecting weak ones, reducing the effects of overfitting.

We demonstrate the efficacy of this approach by utilizing ChIP-chip data of the enhancer-blocking insulator CTCF, which is also known to act as chromatin barrier and to play a fundamental role in nucleosome positioning.

Poster No. 12: Identification of cis-regulatory motifs and modules underlying tissue specific expression

M. Santolini, H. Rouault, P. Maire, V. Hakim
Laboratoire de Physique Statistique, Ecole Normale Supérieure, Paris, France

Cellular differentiation and tissue specification depend in part on the establishment of specific transcriptional programs of gene expression. These programs result from the interpretation of the genomic cis-regulatory information by sequence-specific transcription factors (TFs). Decoding this information in sequenced genomes is a key issue. We recently developed a Bayesian phylogeny-based algorithm to computationally identify the cis-regulatory elements that control gene expression in a set of co-regulated genes. Starting with a small number of Cis-Regulatory Modules (CRMs) in a reference species as a training set, but with no a priori knowledge of the factors acting in trans, the algorithm uses the over-representation and conservation of transcription factor binding sites (TFBSs) among related species to predict Position Weight Matrices (PWMs) of putative regulatory elements and genomic CRMs underlying co-regulation. This method was successfully applied to the gene expression program active in Drosophila melanogaster sensory organ precursor cells (SOPs), a specific type of neural progenitor cells, using a set of known CRMs as training set. Here we report the preliminary results of the method application to vertebrate development. We study muscle differentiation, the main stages of which are under the control of the Homeobox proteins SIX, along with co-factors depending on developmental stage and cellular context. We show that, using promoter regions of genes down-regulated in SIX knocked-out mice embryos, one can recover PWMs for regulatory factors such as SIX or MYOD, a major TF involved in muscle differentiation. Existing ChIP-on-chip and ChIP-seq data for these factors allow us to test the specificity of the predicted PWMs on a genome-wide scale. The combinatorial binding underlying the muscle spatio-temporal development is also investigated using expression data at different developmental stages.

Poster No. 13: Species Diversity from Evolution in a Minimal Predator-Prey Network

Ingo Bathmann, Klaus Pawelzik
Institut f. Theor. Physics, University Bremen, Cognium, Hochschulring 18, D-28359, Germany

In population models with pure competition for a resource typically only the fittest species survives, which is well known as the priciple of competitive exclusion. Therefore, stability of diverse species in an ecological network requires specific structures of the network of mutual species interactions. We were interested in the minimal necessary conditions by which such networks could emerge from Darwinian evolution. We propose a simple predator prey model including dynamical resources in which both, the species-species interactions as well as the species-resource interactions are subject to mutations. The mutations take physical constraints into account which we realize by competition among the interactions. We find that these assumptions are sufficient for explaining the origin and maintenance of biological diversity in non-spatial predator-prey systems. In particular, simulations of our model exhibits species bifurcations with more branches than ground resources, interaction matrices with clear clusters, and a long term growth of diversity. Our results demonstrate that also in well mixed habitats simple evolutionary mechanisms can be sufficient to develop and maintain a high degree of diversity which is for instance exhibited by the plankton in the open ocean.

Poster No. 14: A spiking neuron network model for the emergence for the division of labor in a swarm

Sylvain Chevallier (1), Hélène Paugam-Moisy (1), Michèle Sebag (2)
(1) INRIA-Saclay, France
(2) LRI, Orsay, France

The emergence of synchronized rhythmical activity is at the core of many complex systems, from neural cell assemblies to living insect societies. This work proposes a model for task allocation in ant colonies, where each ant brain is modelled through the competition between two spiking neurons. The dynamic coupling of these neurons implements the individual decision making (mainly foraging or resting) depending on local interactions: Basically, receiving spikes of already foraging ants, an ant decides to forage herself unless the foraging activity in its neighborhood be sufficient. Hence an ant colony is modelled by a large and sparsely connected spiking neuron network where the synaptic connections represent virtual spatial neighborhoods.

This micro-scale model (individual behavior) accounts for the emergence of complex spatio-temporal patterns at the macro-scale (cell assembly). Depending on the parameters of the spatio-temporal coupling, three types of dynamic shapes are observed: An asynchronous individualist) mode [A], a synchronous activity (emergent organization) [B] and finally a periodic synchronous activity [C]. A phase diagram w.r.t. the main two order parameters, reporting the transition between the 3 modes, is displayed and will be discussed.

Poster No. 15: Development of a salt and pepper organization of orientation preference in visual cortical networks

Juan D. Florez W., Fred Wolf
Max Planck Institute for Dynamics and Selforganization, Göttingen

Response characteristics of orientation tuned neurons appear to be similar in the visual cortex of long evolutionary separated mammalian linages [Niell and Stryker 2008]. The spatial arrangements of tuning properties across the cortex, however, show fundamental differences. While in primates and carnivores orientation preference of neurons varies smoothly and progressively, in rodents and lagomorphs it is randomly distributed (salt and pepper arrangement) [Ohki et al 2005]. What causes this dissimilarity is unclear. It has been proposed that in the mouse similarly tuned neurons are selectively connected in segregated subnetworks [Ohki and Reid 2007]. Recent findings however rather indicate that neurons are densely connected independent of orientation tuning [Jia et al 2010]. How can the properties of neighboring neurons be so different if they are connected in a dense network? Here we show that a random arrangement of features can naturally emerge within densely connected networks. We study a model in which the developmental dynamics of orientation preferences is modeled by a Landau equation with nonlocal Gaussian interaction kernel. The model has exact spatially ordered (map) solutions, for which the stability of the stationary maps can be analytically studied. Using this approach we examined how the connectivity patterns from the network determine the final organization of preferences. When the excitation range is shorter than inhibition the attractor states of the system are ordered maps. In contrast, with strong short range inhibition and weak wide range excitation, all ordered map solutions are unstable and the attractor state is an apparently random distribution of orientations. Corroborating those results, numerical simulations show that a random arrangement of orientation preferences develops towards an ordered map or stays disordered depending on the connectivity pattern.

Poster No. 16: Uniform curation of metazoan signaling pathways – resource and tool for experiment design and evaluation

Illés J. Farkas (1), Tamás Korcsmáros (2,3), Dávid Fazekas (2), Máté S. Szalay (3), Petra Rovó (2), Zoltán Spiró (3), Csaba Böde (4), Katalin Lenti (5), Tibor Vellai (2), Péter Csermely (3)
(1) Statistical and Biological Physics Group of the Hungarian Academy of Sciences, Pázmány P. s. 1A, H-1117 Budapest
(2) Department of Genetics, Eötvös University, Pázmány P. s. 1C, H-1117 Budapest
(3) Department of Medical Chemistry, Semmelweis University, PO Box 260, H-1444 Budapest;
(4) Morgan Stanley Hungary Analytics Ltd., Lechner Ö. f. 8, H-1095 Budapest;
(5) Department of Morphology and Physiology, Semmelweis University, Vas u. 17, H-1088 Budapest, Hungary

Signaling pathways, e.g., TGF-ß or Notch, control a large variety of cellular processes, and their defects can cause diseases. Reliable analyses of these pathways need uniform pathway definitions and curation rules applied to all pathways. However, current signaling databases rarely apply such rules leading to an underrepresentation of pathway cross-talks and multi-pathway proteins, i.e., proteins functioning in more than one pathway.

We present SignaLink [1], a signaling resource containing 8 major signaling pathways from C. elegans, D. melanogaster, and humans. Compared to three widely used pathway databases SignaLink contains (in the curated signaling pathways) the (1) highest numbers of signaling proteins and interactions; (2) highest numbers of cross-talks and multi-pathway proteins; (3) above the average number of publications used per pathway. With SignaLink we found that in humans any two of the 8 pathways can cross-talk. We analyzed multi-pathway proteins and cancer-related cross-talks and identified 253 proteins possibly important for drug target discovery [2]. We predicted novel pathway components by combining the signaling networks of the three organisms. SignaLink is available at http://SignaLink.org.

In biomedical research protein functions are often altered, and these changes can strongly influence signaling proteins within a few interaction steps. This can lead to unexpected signaling changes and unwanted phenotypes. To assist experiment design and evaluation, we developed the PathwayLinker web service. It links the queried proteins to signaling pathways through the interactions of the protein-protein and the genetic interaction networks. PathwayLinker is available at http://PathwayLinker.org.

In both studies we found that cellular signaling pathways strongly overlap. We provide several interactive online visualizations of the signaling subnetworks. These visualizations are optimized for, e.g., displaying the network of signaling pathways, viewing the ortholog networks of the three species together, or assisting the planners of experiments by listing all signaling functions in the network neighborhood of a selected protein.

References:

1. Korcsmáros T., et. al. Uniformly curated signaling pathways reveal tissue-specific cross-talks and support drug target discovery. Bioinformatics 26, 2042 (2010).
2. Korcsmáros T., et. al. Signologs: novel signaling pathway components identified by pathway annotation transfer between species. (submitted).

Poster No. 17: An Algebraic Approach to Signaling Cascades of Length n

Elisenda Feliu, Lars N Andersen, Michael Knudsen, Carsten Wiuf
Aarhus University

Classic signaling pathways consist of a cascade of post-translational modification cycles whose activated protein acts as the modifier enzyme in the next cycle. In order to understand the biological importance of this sophisticated signaling mechanism and to predict the behavior of the system, a number of theoretical studies have focused on determining the system’s dynamics and steady-states. In the most discussed case, the MAPK cascade, the presence of bistability and oscillatory behavior has been revealed for different choices of system complexes) are typically imposed on the system in order to simplify the mathematics. This has the consequence that some biological mechanisms, like sequestration effects, are ignored, as uncovered by recent works. Here, assuming only mass-action kinetics and taking the formation of intermediate complexes into account, we show that the steady-states of a signaling cascade are solutions to a system of polynomial equations on the chemical species. We show that for a signaling cascade of length n with one cycle of post-translational modification at each layer, the steady-states arise as the solutions of only one polynomial in one variable. This allows us, for the first time, to conclude that these systems have only one steady-state, which is stable. This is achieved using an induction argument that reflects the modularity of signaling cascades. Our approach makes it possible to study aspects of the system in details. For example, stimulus-response curves arise as inverse functions of an explicit rational function (i.e. quotient of polynomials in one variable), allowing for a theoretical analysis of e.g. the Hill coefficient. Several other rational functions are found that describe the variability of steady-state concentrations; these functions provide a clear picture of how the amounts re-accommodate themselves to variations in initial concentrations. In addition, our approach enables us to explore the parameter space in more details than previously and to divide the space into regions that exhibit different behaviors. Our results rely only on the assumption of mass-action kinetics. The simplicity exposed by the algebraic approach has been the key component in this work and we envisage that it might be useful for analyzing many other post-translational modification systems arising in biological processes.

Poster No. 18: Characterising Properties of Artificial Chemistry Binding Systems

Edward Clark (1,2), Adam Nellis (1,2) ,Simon Hickinbotham (1,2), Mungo Pay (1,3), Susan Stepney (1,2), Tim Clarke (1,3), Peter Young (1,4)
(1) York Centre for Complex Systems Analysis
(2) University of York, Department of Computer Science
(3) University of York, Department of Electronics
(4) University of York, Department of Biology

Artificial Chemistries (AC) hold the key to developing systems that can not only implement networks representing biological reactions, but are also capable of evolving such networks in an unprescribed manner. Tractable ACs are not faithful models of how real world Chemistry works, but are abstract systems which can represent molecules and reactions. The field is still immature, but has potential to have important applications in Biology. For example, discovering genetic control motifs that can be used as design patterns for genetic modification of cell that are more difficult for evolution to remove.

We have taken the first steps towards advancing the field to develop ACs that have properties similar to real world Chemistry. This first step is to provide a method to quantifiably measure the properties of a chemistry, be it real or artificial. Measuring the properties of a subset of real chemistry (amino acid chemistry for example) can then provide a specification for the development of an AC with similar properties, but is both computationally tractable and evolvable.

The work does not only concern itself with modelling networks of chemical reactions in a biological system, but how such networks change over time. Synthetic biologists take organisms whose core networks are robust to evolution and add genetic and regulatory control networks that may be unstable.

Poster No. 19: Modelling the chemical dynamics of autocatalytic networks in a pre-biotic scenario

Varun Giri (1), Sanjay Jain (1,2,3)
(1) Department of Physics and Astrophysics, University of Delhi, Delhi 110007, India
(2) Jawaharlal Nehru Centre for Advanced Scientific Research, Bangalore 560064, India
(3) Santa Fe Institute, 1399 Hyde Park Road, Santa Fe, NM 87501, USA

An important and open question in understanding the origin of life relates to the origin of self assembling and self sustaining organized structures. One of the puzzles is the origin of long molecules such as proteins and RNA molecules starting from much smaller precursors that were abundant on the prebiotic earth, and the build up of the long molecules in sufficient concentrations to seed life. We use a mathematical model based on an artificial algorithmic chemistry to investigate how certain specific chemical species can be selected out of a large set of possible combinations. We start by considering a set of small molecules, called input or food set, that represent simple molecular species that might have been abundantly present around a hydrothermal vent. We then construct a network of chemical reactions amongst these molecules and their reaction products. The chemical dynamics of the molecular populations is simulated as a (large) set of coupled ordinary differential equations. We find that under certain circumstances, autocatalytic sets (ACSs) come to dominate the chemistry in that the concentrations of the molecules belonging to an ACS are much higher than the background. We describe a cascading mechanism by which large and improbable molecules are formed relatively easily in our system, thereby making more plausible the appearance of macromolecules like proteins, RNAs, etc., in pre-biotic settings. We further try to quantify how the topology of the underlying chemistry and dynamical parameters such as rate constants determine whether ACSs dominate or not.

Poster No. 20: MES: a theoretical approach for multi-scale emergence and dynamics

Andrée Ehresmann
Université de Picardie Jules Verne, Amiens, France

The Memory Evolutive Systems, developed with Jean-Paul Vanbremeersch since 1986, propose a (mainly qualitative) model for evolutionary multi-level autonomous systems such as biological, social and cognitive systems. Based on a 'dynamic' category theory, integrating multiple temporalities, MES admit a 'multi-scale' self-organization, modulated by the cooperative/competitive interactions between a net of functional subsystems, the 'coregulators', each operating at its own rhythm as a hybrid system on its 'landscape'. This model gives a global approach to the 'binding problem': how meaningful patterns can take a robust identity of their own, how complex links emerge between them; and how to define the internal 'complexity order' of an element constructed through iteration of such 'complexification' processes. A main result characterizes the property ('degeneracy' or 'multiplicity' principle) ensuring the emergence over time of an intertwined hierarchy of elements of increasing complexity order, in particular of higher cognitive processes in the application MENS to neuro-cognitive systems.

References:

1. "Memory Evolutive Systems: hierarchy, emergence, cognition", by Ehresmann and Vanbremeersch (Elsevier, 2007)

Poster No. 21: In silico Evolution of Early Metabolism

Alexander Ullrich
Bioinformatics, University of Leipzig, Leipzig, Germany

We developed a simulation tool for investigating the evolution of early metabolism, allowing us to speculate on the formation of metabolic pathways from catalyzed chemical reactions and development of characteristic properties. Our model consists of a protocellular entity with a simple RNA-based genetic system and an evolving metabolism of ribozyme catalyzed enzymes that manipulate a rich underlying chemistry. Ensuring an almost open-ended and fairly realistic simulation is crucial for understanding the first steps in metabolic evolution. We show here, how our simulation tool can be helpful in arguing for or against hypotheses on the evolution of metabolic pathways. We demonstrate that seemingly mutually exclusive hypotheses may well be compatible when we take into account that different processes dominate different phases in the evolution of a metabolic system. Our results suggest that forward evolution shapes metabolic network in the very early steps of evolution. In later and more complex stages, enzyme recruitment supersedes forward evolution, keeping a core set of pathways from the early phase.

Poster No. 22: Modelling competitions between Saccharomyces cerevisiae life-history strategies

Thibault Nidelet, Aymé Spor, Leila Tlili, Christine Dillmann, Delphine Sicard, Dominique De Vienne, Olivier C. Martin
UMR 0320/8120 de Génétique Végétale, Gif-sur-Yvette, France

In system biology the choice of the emerging phenotype to study is not trivial. Among diverse phenotypes, growth rate is one of the most commonly studied, partly because it is supposed to be directly related to fitness. However, fitness has other components, also called life-history traits that may play a key role in evolution. This is the case for example of survival. To study the importance of different life-history traits, we have modelled S. cerevisiae population dynamics derived from the classical Lotka-Volterra equations. We have simulated the yeast consumption rate of glucose, glucose transformation into ethanol and energy allocation to biomass and growth rate, as well as mortality related to ethanol toxicity. First we show that with our simple model we can predict population dynamics with high accuracy. Second by using a set of model’s parameters from isolated strains we successfully predicted the outcome of competition betweens pairs of strains. This work shows the importance in system biology of taking into account more than just the population growth rate.

Poster No. 23: Assessing the Optimality Assumption in FBA-based Modeling by Network Structure Perturbation

Georg Basler (1), Sergio Grimbs (1), Joachim Selbig (1), Zoran Nikoloski (2)
(1) Institute of Biochemistry and Biology, University of Potsdam, 14469 Potsdam, Germany
(2) Max-Planck Institute of Molecular-Plant Physiology, 14476 Golm, Germany

Background: Application of Flux Balance Analysis (FBA) has lead to predictive genome-scale computational models. Given a metabolic network, FBA identifies the space of feasible metabolic functions (flux distributions) by imposing governing constraints. Optimization is used under the assumption that evolutionary pressure results in optimal metabolic functions with respect to an objective function. The approach has been extended to determine the effects of gene deletions by identifying a feasible metabolic function closest to the wildtype optimum. Therefore, no further constraints on network operation are necessary, provided the appropriate objective function is chosen.

Objective: We assess the extent to which the principle assumption of optimizing an objective function is applicable across different organism- and tissue-specific metabolic networks. There are three possible ways to verify this assumption: (1) alteration of the optimization function, to reflect biological knowledge, (2) perturbation of the underlying network structure, while satisfying the governing constraints, and (3) combination of the two. Recent studies have tested the optimality assumption by altering the objective function. Here, we assess the validity of the optimality assumption by using network perturbations.

Methods: We employ a novel method for perturbing metabolic networks whereby a compound is represented by a mass vector over chemical elements. Mass balance is guaranteed via the concept of mass equivalence, defined in terms of linear dependence of mass vectors. We first perform systematic perturbation of single and pairs of compounds to obtain biochemically feasible reactions. We then apply FBA on each of the obtained networks with the same objective function.
Results: The results from FBA are used to characterize the performance of a perturbed model in the vicinity of the original network. Here, we consider established organism- and tissue-specific models of Barley seeds and E. coli. We focus on the perturbed networks resulting in a higher value of the objective function to identify the reactions crucial for the increase in the optimal value. The existence of such networks warrants caution in applying the optimality assumption inherent to FBA.

Conclusion: Our findings suggest that systematic network perturbations are a useful tool for obtaining a better understanding of the evolutionary optimization in metabolism.

Poster No. 24: The Transcription Factor Cycle (TFC) Hypothesis and the Genon-Concept

Klaus Scherrer
Inst. J. Monod, CNRS, Paris

The Gene and Genon (GENe-operON) concept (1,2) has emphasised the necessity to conceptually separate genomic product information (the coding triplet sequence) from regulative information. Strictly separating these 2 types of information it is possible to apply Information-Theoretic analysis to both. Regulative cis-programs are based on the sequential alignment of sites in DNA and RNA binding regulative protein or RNA (within RNAi) factors. On an individual mRNA, added to and superposed onto the coding sequence is the cis-program representing, thus, its Genon. The cis-genon picks up an ensemble of factors from trans (its transgenon) by the sequential oligomotifs recognising protein or si/miRNA. By analogy, the pre-mRNA carries the pre-genon program (controlling e.g. differential splicing) and the DNA of a genomic domain its proto-genon controlling chromatin modulation as well as transcription start and extent.

The notion of such programs at DNA and (pre-)mRNA level became most important recently, to interpret post-genomic data bearing on the unexpected extent of the transcribed genome and the size of individual primary transcripts. Most important is the notion that, thus, all CRMs (cis-regulatory modules) of promotor, enhancer and other types are transcribed. The sequence of CRMs way upstream, downstream and in intronic or intergenic positions represents thus a pre-genon; clear-cut phenotypes relate to mutation of such CRMs without affecting the coding sequence. This may allow interpreting in a new manner old data showing that factors that are assembled on DNA may be carried away by the nascent transcript, to exert a function at RNA level.

For a long time it is known that certain factors binding DNA with high affinity and site specificity, as e.g. the large T-antigen of SV-40 and Polyoma virus' as well as factors binding MARs, are found predominantly on pre-mRNA. Some TFs and promoter-binding factors are known to bind RNA also. Based on such facts and general principles of the RNA game, the possibility arises that TFs and other DNA-binding factors might be transferred from chromatin to pre-mRNA or full domain transcripts (FDTs); liberated by RNA processing they may cycle back to the DNA in a balance controlled by relative affinity/specificity. Due to higher affinity the DNA might serve as "entry site" binding, and possibly organising into ensembles factors to be carried away by the nascent RNA. FDTs have two functions: (i) control mRNA synthesis in cis and (ii) bind from trans and store RNA-binding factors controlling gene expression. (1) In interphase cells FDTs produce (an ensemble of) mRNA under the control of the pre-genon programs. (2) Producing some mRNA during mitosis and meiosis (e.g. histone mRNA) they, furthermore, might maintain the organisation of factors fixed on (partially processed) pre-mRNA to be, possibly, reinserted site-specific on "virgin" DNA. In mitosis, these factors might be stored as such, or in an organised manner on not (fully) processed FDTs to be, when new interphase chromosomes form, transferred back to the DNA. This might perpetuate and control euchromatin open for transcription. In meiosis, when the genome is fully transcribed (as observed on lampbrush chromosomes, e.g.) and oocytes full of high Mr RNA, the (partially processed) FDTs with their attached factors might be disposed in sectors of the ooplasm. During cleavage stage they may end up in specific (sets of) cells, feeding specific factors to chromatin of individual cells. This process would correspond to a first stage of determination of later differentiation in space and time.

The TFC concept implies also that mutations of TF sites will affect not gene products per se, but interfere in the systems of regulative networks, which control gene expression, leading to specific phenotypes. Workshop CNRS-MIS, Leipzig, 3-4 Oct 2010

References:

• K. Scherrer and J. Jost (2007) Molecular Systems Biology 3; 87; doi:10.1038/msb4100123
  The gene and the genon concept : A functional and information-theoretic analysis
• K. Scherrer and J. Jost, (2007) Theory Biosci. 126:65–113; DOI 10.1007/s12064-007-0012-x
  Gene and genon concept : coding versus regulation. A conceptual and information - theoretic analysis of genetic storage and expression in the light of modern molecular biology - Essay

Poster No. 25: Sampling Viable Genotypes in Metabolic Reaction Spaces

Areejit Samal (1,2), João F. Matias Rodrigues(3), Jürgen Jost (2), Andreas Wagner (3), Olivier C. Martin (1,4)
(1) Laboratoire de Physique Théorique et Modèles Statistiques, CNRS and Univ Paris-Sud, Orsay, France
(2) Max Planck Institute for Mathematics in the Sciences, Inselstr. 22, 04103 Leipzig, Germany
(3) Department of Biochemistry, University of Zurich, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland
(4) Institut National de la Recherche Agronomique, Laboratoire de Genetique Vegetale du Moulon, Universite Paris-Sud, Gif-sur-Yvette, France

The chemical reactions that an organism can catalyze define its metabolic genotype. If the organism is capable of producing all needed biomass components in a given environment, it is said to be viable. How do genome-scale metabolic networks with a given viability vary in their reaction content? Compared to other systems, little is known about the organization of metabolic phenotypes in metabolic genotype space. Here we study metabolic genotypes whose phenotype is viability in minimal chemical environments that differ in their sole carbon sources. By using Monte Carlo sampling, regardless of the number of reactions in a metabolic genotype, we find that viable genotypes form vast, connected, and unstructured sets -- genotype networks -- that typically nearly span the whole of genotype space. The robustness of metabolic phenotypes to random reaction removal in such spaces has a narrow distribution with a high mean. Different carbon sources differ in the number of metabolic genotypes in their genotype network; this number decreases if a genotype is required to be viable on more than one carbon source, but much less than if the reactions were used independently across different chemical environments. Our work shows that phenotype-preserving genotype networks have generic organizational properties and that these properties are insensitive to the number of reactions in metabolic genotypes.