Speakers' abstracts

Jonas S. Almeida

Automated variable discovery for biomarker identification - how knowledge engineering became the new frontier of machine learning.

Conventional model identification techniques are challenged by the large number of candidate dependent parameters produced by high throughput molecular biology methodologies. That needle in the haystack challenge is met by advanced variable selection techniques. The first results of the improved biomarker identification methodology suggests a second challenge: critical co-variates for complex biological processes are found in distinct omic compartments. For example, there are critical genomic covariates to effective proteomic biomarkers. In other words, discovering the needle in the haystack is compounded by the need to identify the set of haystacks worth searching. The tandem challenge leads to the observation that knowledge representation in a computable form becomes a critical resource for biomarker identification. In this context, the use of the semantic web for knowledge engineering and its articulation of statistical computing is briefly described.

Filipa Alves

Three-dimensional models for cell polarization in pollen tubes.

The germinating pollen is a useful paradigm for apical cell growth and morphogenesis. Different experimental approaches have already produced high quality data, ranging from molecular biology to electrophysiology, imaging and transcriptomics.

We present a theoretical model analyzing the contribution of ion dynamics to cell polarization in pollen tubes. A partial differential equation model incorporating existing knowledge on ion fluxes across the plasma membrane and ion transporters distribution is shown to predict the observed intracellular ion gradients. We will discuss the minimal conditions that are necessary to achieve the pH and Calcium gradients observed in vivo. Furthermore, the proposed model provides a simple rationale for species-specific differences in the intracellular ion gradients based on differences in the magnitude of fluxes and cell size.

Upon extensive experimental validation and iteration of the modeling effort we aim to establish a set of theoretical approaches that can contribute to a better understanding on how ion dynamics may contribute to regulate polarization and apical growth in pollen tubes.

Rui Alves

Bacterial Two Component Systems

Bacterial two-component systems (TCS) are key signal transduction networks regulating global responses to environmental change. Environmental signals may modulate the phosphorylation state of sensor kinases (SK). The phosphorylated SK transfers the phosphate to its cognate response regulator (RR), which causes physiological response to the signal. Frequently, the SK is bifunctional and, when unphosphorylated, it is also capable of dephosphorylating the RR. The phosphatase activity may also be modulated by environmental signals. We constructed mathematical models to examine the steady-state and kinetic properties of the network.

Mathematical modelling reveals that a) Bifunctionality has physiological implications that can justify its selection, and b) The TCS can show bistable behaviour for a given range of parameter values if unphosphorylated SK and RR form a dead-end complex that prevents SK autophosphorylation. Additionally, for bistability to exist the major dephosphorylation flux of the RR must not depend on the unphosphorylated SK. Structural modelling and published affinity studies suggest that the unphosphorylated SK EnvZ and the RR OmpR form a dead-end complex. However, bistability is not possible because the dephosphorylation of OmpR approximately P is mainly done by unphosphorylated EnvZ. Some implications of this potential bistability in the design of a TCS network will be discussed.

Fernado Antunes

The mechanism of cytochrome C oxidase inhibition by nitric oxide

Nitric oxide (NO) emerged as a physiological regulator of the mitochondrial respiratory chain but the detailed molecular mechanism underlining this regulation puzzles investigators. The inhibition of cytochrome c oxidase (CcOX) by NO is a fast process (seconds or faster) observed at high ratios (40 to 500) of O₂ to NO. But paradoxically the known kinetic data seem to favor the complex of CcOX with O₂ over that with NO: binding rate constants of NO and O₂ to the reduced binuclear center Cu_B/a₃ of CcOX are similar; NO dissociates slowly from this center while O₂ is kinetically trapped. In this work, simple mathematical models are applied to investigate this paradox. The results showed that all known features of the inhibition of CcOX by NO can be accounted by a direct competition between NO and O₂ for the reduced binuclear center Cu_B/a₃ of CcOX. Besides conciliating apparently contradictory data, this work provided an explanation for the so-called excess capacity of CcOX, by showing that the CcOX activity found in tissues is optimized to avoid an excessive inhibition of mitochondrial respiration by NO. In pathological situations, like CcOX deficiency diseases and chronic inflammation, an excessive inhibition of the mitochondrial respiration is predicted. Concerning, the reaction of NO with the oxidized binuclear center CuB/a3, whose role remains unknown, we suggest it acts as a safe valve against excessive NO inhibition. Overall, CcOX emerges as a bifunctional enzyme: besides its classic role in the respiratory chain it also acts as an NO oxidase.

David Baker - KEYNOTE SPEAKER

Novel enzymes, rapid structure determination, and a multiplayer computer game for studying folding and design

I will describe the design of novel enzyme activities from scratch using new computational protein design methods. I will also describe recent advances in protein structure prediction which make possible, in favorable cases, the prediction of unknown structures with near atomic accuracy. Finally, I will describe an on-line protein folding and design game (http:/fold.it) for both research and science education.

Antonio M. Baptista

Molecular dynamics simulations at constant reduction potential and pH

Molecular dynamics (MD) simulation methods are one of the major tools in biomolecular modelling. Despite its widespread use, the standard MD methods have several limitations, including the necessity to choose a unique set of protonation and redox states for the protein ionizable sites. This contrasts with the behavior of a real protein in solution, which exhibits binding equilibrium of protons and electrons in a way that reflects the pH and reduction potential of the surrounding solution. From a theoretical viewpoint this real situation corresponds to a semi-grand canonical ensemble, not sampled by standard MD method. We have developed a stochastic method [1] that samples from the proper ensemble, based on a combination of methodologies: standard MD (to sample conformations), Poisson-Boltzmann (to compute ionization free energies) and Monte Carlo (to sample ionization states). This method was initially developed to address protonation events only and was applied to study several pH-dependent processes: helix-coil transition of polylysine [2], conformational classes of kyotorphin [3], acidic titration of lysozyme [4]. Recently, the method was extended to include redox events and was applied to study a multi-heme cytochrome. This talk presents an overview of the theory and its applications, with emphasis on the recent extension to redox processes.

[1] Baptista, A. M., Teixeira, V. H., Soares, C. M. (2002) Constant-pH molecular dynamics using stochastic titration. J. Chem. Phys. 117:4184-4200.

[2] Machuqueiro, M., Baptista, A. M. (2006) Constant-pH molecular dynamics with ionic strength effects: the protonation-conformation coupling in decalysine. J. Phys. Chem. B, 110:2927-2933.

[3] Machuqueiro, M., Baptista, A. M. (2007) The pH-dependent conformational states of kyotorphin: a constant-pH molecular dynamics study. Biophys. J. 92:1836-1845.

[4] Machuqueiro, M., Baptista, A. M. (2008) Acidic range titration of HEWL using a constant-pH molecular dynamics method. Proteins, 72:289-298.

Andreas Bohn

Multi-level modeling of stochastic biomass growth in phototrophic biofilms

The present work deals with the development of stochastic growth models of mixed cultures of hetero- and autotrophic organisms, the so-called phototrophic biofilms. A central objective is to characterize deviations from typical logistic dynamics, induced by randomly occurring detachment events, and to pinpoint their dependence on specific environmental conditions and physiological traits of the biofilms. For this purpose, three different quantitative perspectives are integrated: i) stochastic models predicting the deterministic development of biofilm biomass as well as the frequency and size of detachment events, ii) data-driven models on biofilm physiology emerging from a database generated in the course of a recent European project on phototrophic biofilms, and iii) individual-based models yielding an explicit representation of the spatiotemporal dynamics of phototrophic biofilms. The outcome of this work might stimulate future work in computational ecophysiology by providing i) generic growth models representing dynamical processes on other time- and space scales, ii) computational tools for the management and analysis of heterogeneous and low-abundance data sources, and iii) improved protocols for future interdisciplinary projects. These tools might also serve to catalyze the development of biotechnological innovations, e.g. for nutrient removal in wastewater treatment.

Rui M. M. Brito

Understanding amyloidogenesis in Transthyretin "with a little help from" Computational Biology

In the last 25 years, a series of neurodegenerative diseases have been identified as amyloid diseases. Among these are Alzheimer´s, animal and human spongiform encephalopathies, familial amyloid polyneuropathy (FAP), and several other disorders. In the case of FAP, a hereditary disease with some incidence in Portugal and other foci in Sweden and Japan, the amyloid aggregates and fibrils are mainly formed by variants of Transthyretin (TTR). TTR is a homotetrameric protein present in the blood plasma and cerebral spinal fluid. The formation of amyloid aggregates and fibrils by TTR involves an initial step where the native tetramer dissociates to monomers, and these undergo partial unfolding and aggregation. Recently, using a series of computational methodologies, we have been exploring several issues in TTR amyloidosis. We have explored the unfolding landscape of TTR, searching for molecular intermediates that might be responsible for protein aggregation. We have built a molecular model of an amyloid protofilament of TTR, based on experimental evidences. And now, we are paying particular attention to the rational design of compounds with potential to inhibit TTR dissociation and assembly.

In this talk, we will present highlights of our research in these topics.

Joao Carriço

OSTIS - Online Sequence-based Typing Information System: a flexible online system for sequence based typing methods.

AIM:With the increasing availability and decreasing cost of DNA sequencing technologies, sequence-based typing methods are being preferred over traditional molecular methodologies, due to their results portability and reproducibility. The ability to share the data between laboratories is increased and facilitated with the appearance of centralized online databases. In this work, we present an easily configurable online information system, which allow users to compare their sequence with a set of prototypes by multiple sequence alignment, and submit their data to a public database where a curator can validate and assign a type identifier.

METHOD: The system was implemented in open-source and freely accessible software. A 'LAMP' (Linux Apache MySQL and PHP) software solution stack was used. The system was design to allow three types of users access: anonymous, registered and curator. Anonymous users can compare their data with the online public database; Registered users are allowed to keep a private record of their data and submit their data to the public database. Curators define which strains define new types and verify the registered users submissions, before adding them for the public database with a type identifier assigned to them. The system uses MUSCLE and ClustalW software for sequence alignment and Jalview for displaying the sequence comparisons and clustering trees for selected sequences.

RESULTS: The system was successfully implemented for ccrB typing (www.ccrbtyping.net) of staphylococci sp isolates carrying the mecA gene. This method was developed as a first line SCCmec typing strategy, since the ccrB sequence is part of the ccrAB locus, whose allotypes are used for the definition of SCCmec types. Clustering of ccrB sequences has been shown to properly discriminate between different SCCmec types (I, II , III, IV and VI) (Oliveira et al, J. Antimicrob Chemother 58:23-30). The online information system currently has 17 distinct ccrB allele sequences on the public database and 98 submitted ccrB sequences, of S. aureus, S epidermidis and S. saprophyticus.

CONCLUSION: The OSTIS flexibility allows for the rapid implementation of any sequence-based typing system based on the sequence similarity to a defined set of sequences (such as emm typing for Group A Streptococcus). The user-curator interaction implemented in the system allows for the curator to quickly validate the user submitted data.

Marta Cascante - KEYNOTE SPEAKER

The robust metabolic network adaptations in multifactorial diseases as new targets for novel designed therapies.

Several techniques as DNA sequencing, expression arrays, and proteomic and metabolomic experiments have provided a large amount of new information that cannot be easily interpreted. The integration of all this information in bioinformatic models is likely to be the most interesting tool to understand and to complete an overview picture of the cellular processes. Metabolic profile is the end point of the signaling events, where changes caused by diseases may be reflected. Here we present a strategy where data from incubation with 13C labeled substrates and the later isotopomer analysis of selected metabolites, are analyzed using a software developed in our laboratory to estimate dynamic flux distribution among the metabolic network. This metabolic status could be correlated with data from other -omics. We identify the main steps that control a metabolic pathway, which may be used as new therapeutical targets. By using this strategy we have identified that maintenance of oxidative and nonoxidative pentose phosphate pathways unbalance is critical for cancer cell survival and vulnerable to chemotherapeutic intervention. Moreover, we have used Metabolic Control Analysis (MCA) to identify the main enzymes controlling ribose-5-phosphate synthesis as well as to design combined target strategies, which have been validated by using specific inhibitors. This strategy results of great interest for the study of other multifactorial diseases. In particular, it is being used to achieve a better understanding of glucose metabolic network to design interventions at a metabolic level in chronic obstructive pulmonary disease (COPD). This new principle for rational drug design has its origin from the integrative, systems biology approach of understanding cell function and opens new ways to develop novel treatments for multifactorial diseases.

This work was supported by funds of Spanish Government and European Union FEDER SAF2005-01627 ISCIII-RTICC (RD06/0020/0046); European Comission (FP6) BIOBRIDGE LSHG-CT-2006-037939; Comissió d'Universitats i Recerca de la Generalitat de Catalunya (2005SGR00204).

Claudine Chaouiya

New means for the qualitative analysis of large regulatory networks

The control of essential cellular processes is performed by complex regulatory networks. The understanding of such complex systems requires the use of dedicated and efficient mathematical modelling frameworks. In this context, the logical formalism allows a qualitative description of regulatory networks and the analysis of their dynamical features. However, even in this discrete framework, one faces a classical combinatorial explosion that calls for the development of new methodological means. Two strategies to tackle large regulatory networks will be presented. On the one hand, the analysis of regulatory circuits provides a useful tool. Indeed, regulatory circuits are known to underlie the emergence of dynamical properties: positive circuits are involved in the existence of multiple differentiated states, whereas negative circuits can lead to cyclical behaviours. On the other hand, we have recently defined a method for model reduction that preserves the main dynamical properties of the system. This reduction allows the consideration of a full-fledged model yet providing a reduced version amenable for dynamical analyses. The circuit analysis and reduction techniques will be briefly illustrated through a model of T lymphocytes activation and differentiation.

Lounes Chikki

Conservation genetics of orang-utans: can we separate demography from population structure.

Habitat destruction and fragmentation represent one of the major threats to biodiversity, and this is particularly important for large mammals which often require large habitats. In Borneo, many primary forests have been replaced by oil-palm plantation amd orang-utans survive in often small and isolated forest fragments. This is particularly true in the Sabah state, in northern Borneo. Using a relatively simple demographical model and a Bayesian coalescent-based inference method, we have shown that genetic data can be used to determine whether (i) a signal for a populaiton bottleneck could be detected, (ii) we could quantify and date the change in population size, and (iii) it would be possible to relate this date to natural or anthropogenic changes. Since that work, we have also tried to quantify the extent to which population structure can mimic population size changes, using simulated data sets. In this talk I shall present the general background of the study, the original results and the new simulation-based results.

Francisco Couto

Untangling BioOntologies for mining biomedical information.

Biomedical research generates a vast amount of information that is ultimately stored in scientific publications or in databases. The information in scientific texts is unstructured and thus hard to access, whereas the information in databases, although more accessible, often lacks in contextualization. The integration of information from these two kinds of sources is crucial for managing and extracting knowledge. By structuring and defining the concepts and relationships within a biomedical domain, BioOntologies have taken a key role in this integration. This talk will describe the role of BioOntologies in sharing, integrating and mining biological information, discuss some of the most relevant BioOntologies and illustrate how they are being used by automatic tools to improve our understanding of life.

Francisco Dionisio

Evolution of spite and biological warfare

Biological warfare, also known as a germ warfare or biological weapons, is the use of any pathogen or parasite as a means of harming other individuals. The conditions for individuals to use their pathogens/parasites as biological weapons to harm other individuals of the same species are: 1) individuals should live in a structured environment; 2) biological weapons should hit mostly non-relatives to the harming individual; 3) and the cost of infecting others should be much lower than the cost of being infected. These three conditions are easily fulfilled by most host-parasite dyads in nature. Mathematical conditions for the spread of spiteful individuals will be given.

Patricia Faisca

Identifying critical residues in protein folding: Insights from phi-value and Pfold analysis

We apply a simulational proxy of the phi-value analysis and perform extensive mutagenesis experiments to identify the nucleating residues in the folding 'reactions' of two small lattice Go polymers with different native geometries. These results are compared with data obtained from an accurate analysis based on the reaction coordinate folding probability Pfold, and on the use of structural clustering methods. Our study reveals a complex picture of the transition state. Indeed, for both protein models, the TS is rather heterogeneous and splits-up into structurally different populations. For the more complex topology the identified sub-populations are actually structurally disjoint, which indicates two distinct transition states associated with parallel folding pathways. For the less complex native geometry we found a broad transition state with microscopic heterogeneity. These findings suggest that the existence of multiple transition states may be linked to the geometric complexity of the native structure. For both geometries, the identification of the FN via the accurate Pfold analysis agrees with the identification of the FN carried out with the phi-value analysis. For the most complex geometry, however, the Pfold and phi-value analysis give more consistent results than for the more local geometry. The study of the TS' structure reveals that the FN's residues are not necessarily fully native in the TS. Indeed, it is only for the moste complex geometry, which shows a distinctively larger fraction of long-range bonds, that the traditional interpretation of phi-values provides a more reliable picture of the structure of the transition state. We conclude that the phi-value correlates with the change in folding rate induced by mutation, rather than with the degree of nativeness of the TS. Overall, our results suggest that native folds having predominantly non-local bonds are more suitable targets for phi-value analysis than other protein geometries.

Ana Teresa Freitas

Methods for the analysis of the state spaces of biological networks

Accurate modelling of biological systems requires an interaction model, that includes all the components and how they connect to each other. It also requires the knowledge of the parameters that make it possible to make quantitative predictions about the dynamic behaviour of the system.

A number of formalisms have been proposed that sit in between the fully continuous differential equations model (too complex to be useful) and the exceedingly simple discrete Boolean model (not rich enough). Amongst these formalisms, Piecewise Affine Differential Equation (PADE) are able to model complex dynamics while preserving the simplicity required to make them useful in practice.

There exists a relationship between PADE models and state transition graphs (STG) that can be formalized by computing the discrete state space equivalence of the possible transitions in the continuous domain. The resulting STG is invariant for large ranges of parameter values which can be expressed in the form of inequality constraints.

The STG for networks with more than fifteen genes usually consist of thousands of states, which make them too large to be analyzed by inspection. Methods for querying and validating formal models have recently been tried in systems biology with interesting results. Formal verification based on model checking provides a powerful tool to query models of cellular interaction networks. The problem of posing relevant questions is critical in modelling in general, but even more so in the context of applying formal verification techniques, due to the fact that it is not easy for non-experts to formulate queries in temporal logic. Initial work in this direction has revealed a lack of powerful, high-level languages for specifying biologically-relevant properties.

In order to deal with this problem, we propose the use of patterns, that is, high-level query templates that capture recurring biological questions and that can be automatically translated into temporal logic. The applicability of the developed set of patterns has been investigated by the analysis of an extended model of the network of global regulators controlling the carbon starvation response in Escherichia coli.

Gabriela Gomes

Population heterogeneities and disease incidence

Simple mathematical models based on population averages provide excellent means to investigate macroscopic behavior in infectious disease epidemiology, such as to compare measures of disease transmission across populations, to associate disease shifts at the population level with characteristics of the history of infection at the individual level, or to discover target thresholds for interventions. Their limitations are noticeable, however, when we try to reproduce epidemiological data with accuracy. I will present the results of a series of models to demonstrate that different forms of population heterogeneity can significantly lower endemic and epidemic levels of infection, while the qualitative behavior remains unchanged.

Isabel Gordo

On the advantage of sex and the level of epistasis

The evolutionary advantage of sexual reproduction has been considered as one of the most pressing questions in Evolutionary Biology. While a pluralistic view of the evolution of sex and recombination has been suggested by some, here we model the simplest possible scenario with hope that it may apply to a multitude of species. Real populations are finite and also subject to recurrent deleterious mutations. With these basic model ingredients we show that the maximum advantage of recombination occurs for an intermediate value of the deleterious effect of mutations. Furthermore we show that the conditions under which the biggest advantage of sex is achieved are those that produce the fastest fitness decline when sex is absent (the operation of Muller's ratchet). The quantification of the range of selective effects that favor recombination leads us to suggest that a connection between sex and stress should be expected, as it is found in several species. Contrary to other models that assume the existence of pervasive negative epistasis between deleterious alleles, such assumption is not required and not supported by the data we recently gathered in the model organism Escherichia coli.

Jose Pereira Leal

Intracellular lifestyles and the nature of the minimal genome

Gene loss is one of the forces shaping genome evolution, but has received far less attention than other forces such as duplication or lateral gene transfer. Intracellular parasites and endosymbiontes are extreme cases of gene loss, and thus ideal models to study this process. We have recently found evidence for a selective drive to maintain protein family diversity at the expense of copy redundancy in the reductive genome evolution of intracellular parasites. Surprisingly we observed that there is little overlap between the remaining protein family repertoires of the different reduced genomes, contrasting with the prevailing notion of a minimal set of genes necessary for (parasitic/endosymbiotic) life. Instead, our results are consistent with a random gene loss scenario, mostly unconstrained by any functional requirements, and where the major selective pressure is maintain diversity of family repertoires. In my talk I will discuss these results and their implication to the notion of minimal gene sets and minimal genomes.

Paulo Martel

Molecular dynamics and docking studies of ligands to Cytochrome P450 3A4

Cytochromes P450 (CYPs) are a large family of oxidase enzymes present in all groups of living organisms. They play a important role as detoxifying enzymes, metabolizing the chemical modification of many foreign substances (xenobiotics) in order to facilitate their elimination from the body. Pharmaceutical drugs are one class of xenobiotics whose metabolic fate depends upon CYP activity, the half-life of a drug in the body thus depending on the particular activity of the CYP that metabolizes it. If a drug is rapidly transformed it can be rendered useless, while a slow transformation can result in potentially lethal high blood levels. Natural genetic variability of CYP enzymes occurs in human populations, with different individuals displaying different variants in the genetic sequence of their CYP enzymes. Understanding how this genetic variability affects enzyme activity is very important because it will allow drug dosage prescription on an individual basis.

One of the central P450 enzymes in human drug metabolism is CYP3A4, which is responsible for the hepatic detoxification of a vast family of foreign substances. The ability of CYP3A4 to metabolize such a diverse array of compounds, many of them bearing little resemblance, stems from its large and flexible active site. Because of these unique characteristics the computer modeling of ligand docking to CYP 3A4 is a very difficult task. We have used a combination of molecular dynamics simulations and computer-assisted docking to investigate the binding mode of various ligands of pharmacological interest to CYP3A4.

Denis Noble - KEYNOTE SPEAKER

The principles of Computational Systems Biology, illustrated with the virtual heart.

Computational physiology (Crampin et al., 2004) is an essential component of the rapidly-growing field of Systems Biology. The Physiome Project has an important contribution to make since it works at all levels of biological organisation. In this lecture I will outline some principles of Systems Biology (Noble 2008), illustrated with examples from computational biology of the heart (Noble, 2006).

Crampin, E.J., Halstead, M., Hunter, P.J., Nielsen, P., Noble, D., Smith, N. & Tawhai, M. (2004) Computational Physiology and the Physiome Project, Experimental Physiology, 89(1), pp. 1-26.

Noble, D. (2006) The Music of Life, OUP.

Noble, D. (2008) Claude Bernard, the first systems biologist, and the future of physiology. Experimental Physiology, 93, 16-26.

Ana Nunes

Stochastic effects in infection dynamics

During the last decade, more sophisticated approaches building on the traditional SIR and SEIR models have brought considerable advances in understanding and selecting some of the fundamental ingredients of the complex dynamics of infectious diseases.

This body of work belongs to an essentially deterministic framework, where demographic stochasticity plays a secondary role, that of sustaining small amplitude fluctuations around the deterministic system's equilibrium that follow the natural frequency given by the local linear approximation.

We show that the breakdown of the assumptions of random mixing of the population and/or of constant recovery rate during the infectious period implies important corrections to the amplitude and dominant frequency of these stochastic fluctuations.

The finding that in finite, discrete populations internal noise together with correlations produces sustained incidence oscillations of significant amplitude all over the parameter region that includes childhood infectious diseases is of importance for the long-standing controversy in epidemiology and ecology as to the driving mechanisms of the pervasive noisy oscillations observed in these systems.

Arlindo Oliveira

New Combinatorial Methods for the Detection of Biologically Significant Motifs in Promoter Sequences

The identification of biologically significant motifs in promoter sequences plays an important role in the discovery and understanding of gene regulatory networks. A large number of algorithms for identifying regulatory regions and motifs have been proposed in the literature. The most commonly used methods to date are based on probabilistic methods that are effective in the identification of strong signals but that are less useful in the identification of weaker signals. Combinatorial methods, based on the systematic enumeration of over-represented sequences of bases have not been extensively used, because they may generate long lists of candidate motifs. In this talk, I will focus on recently developed methods based on combinatorial algorithms that have the potential to become the tool of choice for researchers interested in identifying motifs that could correspond to transcription factor binding sites in the promoter regions of target genes.

Jose Luis Oliveira

Extraction and integration of information from biomedical databases

Biomedical information is currently spread along a large set of public and private data repositories. However given its different formats, contents, access methods and, particularly, the set of interactions between biological and medical concepts it is not easy to extract knowledge from all that data.

The bioinformatics group at the University of Aveiro has been developing several techniques for extracting, integrating and visualize genotype-to-phenotype data. GeneBrowser, for instance, is a web portal that can be used to better understand the biological relations between a set of genes. It integrates bibliography information with functional annotations, using Gene Ontology, Entrez Gene, KEGG Orthology and KEGG Pathways.

(More information on http:// bioinformatics.ua.pt/)

Jorge Pacheco

Evolutionary games on self-organizing networks

I will discuss the evolutionary dynamics of populations in which individuals engage in games associated with popular social dilemmas. The dynamical structure of their social ties co-evolves with individual strategies, such that individuals differ in the rate at which they seek new interactions. Moreover, once a link between two individuals has formed, the productivity of this link is evaluated. Links can be broken off at different rates. Whenever the active dynamics of links is sufficiently fast, population structure leads to a transformation of the payoff matrix of the original game. We explore the evolutionary dynamics of one shot games, deriving analytical conditions for evolutionary stability.

Francisco Pinto

The biological meaning of graph based node similarity measures

Graph based similarity measures as GTOM (Generalized Topological Overlap Measure) have been used with success in the partition of biological networks into functional modules that allow a deeper understanding of the overall organization of the referred networks. We have developed a probabilistic model of the pairwise response of genes to random network perturbations and found that the former graph based measures correspond to a composition of the probabilities with which two genes respond simultaneously or differentially to random network perturbations. Therefore, we propose that the success in the definition of functional network modules is due to the biological meaning of the graph based measures.

Maria Joăo Ramos

Computational Strategies in drug design

This talk is concerned with computational strategies devised with drug design in mind. Some fundamental stages of rational drug design are addressed [1], namely the atomic level understanding on disease-related enzymatic mechanisms and inhibition [2,3,4], which can be regarded as a pre-requisite to any attempt to rationally design new, better inhibitors; computational alanine-scanning mutagenesis of protein-protein interfacial residues, which can be a very important process for drug design since protein-protein interactions form the basis for most biological processes [5,6]; and molecular docking using total flexibility of ligand and receptor [7]. All methods described here offer interesting approaches to rational drug design.

[1] M. J. Ramos, P. A. Fernandes, Curr. Comput.-Aided Drug Design, 2, 57 (2006)

[2] S. Pereira, P. A. Fernandes, M. J. Ramos, J. Am. Chem. Soc., 127, 5174 (2005)

[3] N. M. F. S. A. Cerqueira, P. A. Fernandes, M. J. Ramos, Chemistry Eur. J., 13, 8507 (2007)

[4] M. J. Ramos, P. A. Fernandes, Acc. Chem. Research, DOI: 10.1021/ar7001045 (2008)

[5] I. S. Moreira, P. A. Fernandes, M. J. Ramos, J. Comp. Chem., 28, 644 (2007)

[6] I.S. Moreira, P.A. Fernandes, M.J. Ramos, Proteins, 68, 803 (2007)

[7] N. M. F. S. A. Cerqueira, P. A. Fernandes, M. J. Ramos, Proteins, in press (2008)

Luis Rocha

Complex Networks and Bibliome Informatics: Applications to Computational Biology

Miguel Rocha

Nature-inspired Metaheuristics for the Optimization of Microbial Strains

In Metabolic Engineering it is difficult to identify which set of genetic manipulations will result in a microbial strain that achieves a desired production goal, due to the complexity of the metabolic and regulatory networks and to the lack of appropriate modelling and optimization tools. In this talk, we report on our recent work where we propose new algorithms (Evolutionary Algorithms and Simulated Annealing) to reach a good set of gene deletions to apply to a given microorganism, in order to maximize a given objective function. In this approach, each mutant strain is evaluated by resorting to the simulation of its phenotype using methods that rely on steady-state assumptions (such as Flux-Balance analysis).

Armindo Salvador

Moiety-supply units in metabolic networks: canonic design principles versus actual realizations

Moiety-transfer cycles whereby a pair of coupled reactions transfers a molecular part (moiety) from a donor to an acceptor molecule via a cycled intermediate are among the simplest and most prevalent elementary circuits in metabolic networks. Most of these cycles occur at the interface between catabolism and anabolism, and play a role analogous to that of power-supply units in electronic circuits. Namely, they must reliably supply moiety at the required rate (analogous to current intensity) while keeping the concentration of the charged carrier (analogous to electric potential) fairly constant. In this talk I will discuss a set of design principles that might apply to moiety-supply units as a general class of metabolic circuits and examine the extent to which concrete biological realizations adhere to those design principles.

Joao Aires de Sousa

Data mining a genome-scale database of metabolic reactions

The understanding of the multiple ways small molecules affect biological systems is demanding an integration of chemical and biological data in 'systems biology' approaches. In this context, 'enzymatic function' emerges as a key gateway between chemoinformatics and bioinformatics. Employed for the annotation of genes and proteins, or encoded as edge in metabolic networks graphs, 'enzymatic function' has an intrinsic chemical nature - it is the catalysis of a chemical reaction.

The automatic perception of chemical similarities between enzymatic reactions is required for a variety of applications from the validation of reaction classification systems, to genome-scale reconstruction (or comparison) of metabolic pathways, to the classification of enzymatic mechanisms. Processing chemical reactions by statistical or machine learning techniques requires a suitable representation. We have proposed the MOLMAP reaction descriptors to numerically encode the structural transformations resulting from a chemical reaction (J. Chem. Inf. Model. 2005, 45, 1775-1783; Angew. Chem. Int. Ed. 2006, 45, 2066-2069).

This talk presents the application of the MOLMAP descriptors to a) the mapping of a genome-scale database of metabolic reactions by Kohonen self-organizing maps leading to the identification of similar reactions bearing differences at the class level of the official EC (Enzyme Commission) number; b) the training of Random Forests that automatically assign EC numbers from the reaction equation; c) the classification of metabolic pathways. Pathways could be clustered according to the type of metabolism. Similarities between pathways could be perceived based on reactions similarities, and making no use of EC numbers.

Andre Valente

Functional Organization of the Yeast Proteome by a Novel Yeast Interactome Map

It is hoped that comprehensive mapping of protein physical interactions will facilitate novel insights, regarding both fundamental cell biology processes and the pathology of diseases. To fulfill this, finding good solutions to two issues will prove essential: i) how to obtain reliable interaction data in a high-throughput setting and ii) how to structure interaction data in a meaningful form, amenable and valuable for further biological research. In this article, we structure an interactome in terms of predicted permanent protein complexes and predicted transient, non-generic, interactions between these complexes. The interactome is generated via an associated novel computational algorithm, from raw high-throughput affinity purification/mass spectrometric interaction data. We apply our technique to the construction of a new interactome for S cerevisiae, showing it yields reliability typical of low-throughput experiments, out of high-throughput data. We discuss relevant biological questions in the context of this novel interactome including, via homology, how it relates to human disease.

Susana Vinga

Metabolic network structure analysis: a case study on glycolysis in Lactococcus lactis

The dynamic modeling of metabolic networks constitutes a major challenge in systems biology. The time evolution of metabolite concentration in cells is modeled by complex systems of non-linear differential equations with a large number of parameters. A case study of the stability analysis based on the Jacobian matrix of the model equations combined with singular value decomposition of output sensitivities is presented for glycolysis in L. lactis. This approach shows how a preliminary structural model can be reformulated in simplified form to substantially improve the parameter estimation task.