EN FR
EN FR


Section: New Results

Data integration

Participants : Jacques Nicolas, Andres Aravena, Charles Bettembourg, Jérémie Bourdon, Jeanne Cambefort, Guillaume Collet, Olivier Dameron, Damien Eveillard, Julie Laniau, Sylvain Prigent, Anne Siegel, Sven Thiele, Valentin Wucher.

Metabolic network reconstruction: combinatorial gap-filling method We introduced an exhaustive gap-filling procedure on the first metabolic network for a macroalgea (Ectocarpus Siliculosus). As this species is a non benchmark model, this issue is related to hard combinatorial optimization problems. To that matter, we took advantages of the latest improvement of Answer Set Programming solvers (combination of clasp and unclasp) and introduced a new model of the network expansion problem. [G. Collet, D. Eveillard, S. Prigent, A. Siegel, S. Thiele] [27]

Identification of functional gene units in non benchmark models We introduced the concept of "shortest genome segments" (SGS) to detect functional units on exotic species, such as extremophiles, that are by nature unrefined. They correspond to genome portion which contain a large density of genes coding for enzymes which regulate successive reactions of metabolic pathways. There identification is a hard optimization combinatorial problem. We relied on the declarative modeling power of answer set programming (ASP) to encode the identification of shortest genome segments and prove that SGS are stable in (i) computational time and (ii) ability to predict functional units when one deteriorates the biological knowledge [D. Eveillard, A. Siegel, S. Thiele] [26]

Refinement of regulatory network from genomic, expression data and functional unit data We integrated heterogeneous information from two types of network predictions to determine a causal explanation for the observed gene co-expression. We modeled this integration as a combinatorial optimization problem. We demonstrated that this problem belongs to the NP-hard complexity class. We proposed an heuristic approach to have an approximate solution in a practical execution time. Our evaluation showed that the E.coli regulatory network resulting from the application of this method has higher accuracy than the putative one built with traditional tools. Applications to the mining bacterium Acidithiobacillus ferrooxidans allowed analyzing the relevance of central regulators. [A. Aravena, D. Eveillard, A. Siegel] [23] , [13] [Thesis]

Reconstruction of a protein interaction network for archaebacteria To gain insights into genomic maintenance processes in hyperthermophilic archaea, a protein-interaction network centered on informational processes of Pyrococcus abyssi was generated by affinity purification coupled with mass spectrometry. We have proposed a graph theoretic analysis of this network including statistical (e.g. clusterisation coefficients) and topological aspects (bicluster analysis, search of a maximal interaction skeleton), which helps network interpretation in terms of formation of complexes or interaction dynamics. [J. Nicolas] [20] [Online publication]

Knowledge evolution in ontologies We studied the impact of an ontology evolution on its structural complexity. As a case study we used sixty monthly releases of the Gene Ontology and its three independent branches i.e. biological processes (BP), cellular components (CC) and molecular functions (MF). For each release, we measured complexity by computing metrics related to the size, the nodes connectivity and the hierarchical structure. We showed that the variation of the number of classes and relations in an ontology does not provide enough information about the evolution of its complexity. However, connectivity and hierarchy-related metrics revealed different patterns of values as well as of evolution for the three branches of the Gene Ontology [O. Dameron, C. Bettembourg] [17] , [14] [Online publication] [Thesis]

Treatment process representation for breast cancer patients. The general cancer registry of Poitou-Charentes developed a multiple source information system covering diseases, anatomical structures and cytopathology. We proposed an algorithm for representing and analyzing the patient's treatment process. An expert compared the original data with our representation and computed a score of dissimilarity. The results showed that an integrated information system can successfully analyze the data to determine whether they comply with the guidelines [O. Dameron] [31] .

AphidAtlas project We began a collaboration with the AphidAtlas project for defining the structure of an ontology of aphids anatomy and development [O. Dameron] [30] .