Méthodes qualitatives pour la construction et l'analyse des réseaux moléculaires SBGN

Abstract : Two fundamental tasks of Systems Biology are the construction of molecular networks from experimental data, and their analysis with a view to discovering their emergent properties. With the increase of available experimental data, these two tasks can no longer be realized by hand. Based on this observation, numerous bioinformatics methods aiming at the automation of these two task have been developped.In parallel, standards aiming at defining and organizing terms of systems biology, or representing networks and mathematical models, have been developped. Among these standards, the Standard Biology Graphical Notation is composed of three languages that allow the representation of molecular networks. The two main SBGN languages are SBGN-PD for the representation of reaction networks, and SBGN-AF for the representation of influence graphs. The SBGN notation not only standardizes the representation of networks, but also gives the concepts of systems biology that are most often used to express knowledge of the field.Our work takes its root in this general background. We have developped a number of methods to construct molecular networks and analyze their dynamics. All the methods that we propose are based on qualitative formalisms, such as logics or automata networks. These formalisms have solid theoretical bases and can be used by numerous pieces of software. All our methods also rely on the biological concepts given by the SBGN standard, and can therefore be blended in the same theoretical framework.First, we introduce two sets of predicates that allow to translate any SBGN-PD or SBGN-AF network into a set of ground atoms. Then, we show how these sets of predicates can be used to reason on networks, by proposing a transformation method of SBGN-PD signaling networks into SBGN-AF influence graphs.Second, we present a first-order logic based method to construct signaling networks from experimental results. This method formalizes and automatizes biologists' reasoning using explicit reasoning rules.On the contrary to existing methods, it allows to take into account numerous types of experimental results while reconstructing precise molecular mecanisms.Third, we show a new method to compute the finite traces and attractor points of Boolean networks that model SBGN-AF networks and that are parameterized using general principles.Finally, we introduce two new qualitative semantics for the computation of the dynamics of SBGN-PD reaction networks. These semantics are expressed using automata networks. The first semantics extends the classical Boolean semantics by taking into account inhibitions. As to the second one, it relies on the concept of story which introduces a new point of view on reaction networks. Indeed, it allows to model different physical states of the same molecular entity using a unique variable.All the methods that we have developped show how qualitative formalisms can be used to reason on the relations represented by molecular networks in order to discorver new knowledge in systems biology.
Document type :
Liste complète des métadonnées

Cited literature [144 references]  Display  Hide  Download

Contributor : Abes Star <>
Submitted on : Thursday, November 3, 2016 - 1:13:05 PM
Last modification on : Tuesday, April 16, 2019 - 9:09:29 AM
Document(s) archivé(s) le : Saturday, February 4, 2017 - 1:25:51 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01391465, version 1


Adrien Rougny. Méthodes qualitatives pour la construction et l'analyse des réseaux moléculaires SBGN. Bio-informatique [q-bio.QM]. Université Paris-Saclay, 2016. Français. ⟨NNT : 2016SACLS325⟩. ⟨tel-01391465⟩



Record views


Files downloads