Section: New Results

Systems Biology

Reasoning on knowledge to build signaling networks:

We have introduced a logic-based method to infer molecular networks and show how it allows inferring signalling networks from the design of a knowledge base. Provenance of inferred data has been carefully collected, allowing quality evaluation. Our method (i) takes into account various kinds of biological experiments and their origin; (ii) mimics the scientist's reasoning within a first-order logic setting; (iii) specifies precisely the kind of interaction between the molecules; (iv) provides the user with the provenance of each interaction; (v) automatically builds and draws the inferred network  [29] .

Metabolic pathways

The topological analyse of metabolic networks is a first step to understand their behaviours and is described in term of fluxes analyses. We work on the elaboration of a stoichiometric model of Bacillus subtilis where its fluxes analyse predicted transcriptional regulation to be more important for the dynamics induced by glucose than by malate [7] .

In metabolic pathway analyses, the metabolic networks are described in term of biochemical reactions and metabolites. The integration of structural data is required for a comprehensive understanding of the metabolic networks. We represent the metabolic networks with the functional connectivity between the protein functional domains to make more relevant analyses. We used Bio$\Psi$, a formal multi-level description based on elementary actions, to assign functions on structural domains and the elementary flux modes theory to check if the already known pathways remain presents and to identify new ones.

A new version of the software has Mpas (Metabolic Pathway Analyser Software) been developed during a Master2'internship by Gh. Fievet. Meanwhile we have also introduced in the landscape of the cell its membranes and the numerous pumps that facilitate ions transfers, hence taking into account the pH of the cytoplasm, a parameter that fits the cell mytosis cycle and which proves to separate the cancerous/normal status of cells [22] . We now aim at study larger and more elaborate metabolic systems, including the Krebs cycle and the mitochondria influence, thus enhancing the scalability of our method [17] .

Bacterial phenotypic adaptation

We attempt to re-interpret a major event, the initiation of chromosome replication in Escherichia coli, in the light of scales of equilibria. This entails thinking in terms of hyperstructures as responsible for intensity sensing and quantity sensing and how this sensing might help explain the role of the DnaA protein in initiation of replication. We outline experiments and an automaton approach to the cell cycle that should test and refine the scales concept [19] .

Another possible direction to study the mechanisms used by cells to integrate and respond to their environment is to search for a link between two large hyperstructures: the cytoskeleton and the general metabolic activity of the cell. There is extensive evidence for the interaction of metabolic enzymes with the eukaryotic cytoskeleton. We state the hypothesis that the cytoskeleton senses and integrates the general metabolic activity of the cell. The physical and chemical effects arising from metabolic sensing by the cytoskeleton would have major consequences on cell shape, dynamics and cell cycle progression. The hypothesis provides a framework that helps the significance of the enzyme-decorated cytoskeleton be determined [18] .

In order to test these hypotheses, we have added many features to the Hsim simulation software. The main addition being a way to get both the power of expression of the "entity-centred" paradigm and the computational efficiency of global methods, such that Gillespie-like stochastic simulation algorithm (SSA). To achieve this, we have implemented two new algorithms. The first one concerns the possibility to take into account the interactions between two classes of molecules: the one we want to follow the spatial location over time (entities) and the one for which only the evolution of the number of copies over time is relevant.

The second algorithm is an enhancement of the tau-leap variant of the exact Gillespie SSA; This allows to take into account the interactions between globally treated molecules. The Hsim -SSA algorithm performs an adaptive processing of the number of reactions which may have been triggered during the time step. At each time step, the fast reactions are averaged while the slow reactions are fully stochastically treated. This allows Hsim -SSA to be more than 10 times faster than the other tau-leap SSA implementations [28] .

Use of bacteria for biotechnology

Another center of interest has been to find a way to use bacteria as a mean to help us to engineer new biomolecules with specific characteristics. It is sometimes speculated that the equivalent of the polymerase chain reaction might be developed for identification of peptides, proteins or other molecules. Natural amplification systems do exist as in the case of certain autoinducer systems in bacteria. We have been outlined a possible, generic method, the mimic chain reaction, for obtaining peptides with 3-D structures that mimic the 3-D structure of their targets. These targets would include a variety of molecules, including proteins. There are therefore two categories of applications: the ability via amplification firstly to detect a known protein or other target at an extremely low concentration and secondly to obtain a set of peptides that mimic the structure of an unknown target and that can be used to obtain a photofit [20] .