In this thesis I investigate the role of spatial structure, cell-cell interactions and horizontal gene transfer on the genetic composition of growing microbial colonies. In the first part I study how the roughness of the growing layer of the colony depends on the shape of colony-forming cells. To study its impact I develop an off-lattice Eden-like model in which cells are represented as spherocylinders with a variable aspect ratio. I show that the roughness of the expansion front is not significantly affected by the shape of cells and that the dynamic scaling of growing front belongs to the KPZ universality class. Roughness is an important and easy to measure feature which affects the probability of fixation of genetic lineages in the colony. Another feature contributing to the genetic composition of a microbial community is horizontal gene transfer, which is investigated in the second part of this thesis. I develop an agent-based computational model of bacterial cells which grow, divide, and interact mechanically. I focus on plasmid conjugation, in which donors transfer a plasmid (a small, circular DNA molecule) to plasmid-free recipients. I show that bacteria in the expanding colony segregate into sectors of donors and acceptors. Donor sectors grow at the expense of acceptor sectors and that effect can be effectively described by coalescing random walkers that perform biased random walk on the colony expansion front. I use numerical and analytical methods to show that the plasmid eventually spreads to the whole colony given enough time, and I also show that this time is unrealistically long for experimentally determined conjugation rates and therefore real colonies are expected to have both acceptor and donor sectors. Furthermore, my simulations show that segregative plasmid loss at the moment of cell division can counteract the effect of conjugation and can lead to fixation of plasmid free cells. I also show that changes in nutrient concentration and the resultant change in roughness of the expansion front affect the rate of plasmid spread into population. Quantitative and qualitative results obtained in this section may serve as a tool to extract plasmid invasion rates from experimental data. In the last part of this thesis I investigate how the physical factors, such as finite strength of conjugative junctions, affect the conjugation process. I develop a computational model of plasmid transfer in which conjugative junctions are explicitly modelled as short, spring-like tubes that connect conjugating cells. My results show that factors such as junction creation rate and its strength can significantly affect the conjugation performance. I study different situations corresponding to different experimental scenarios (well-mixed colony on a filter paper, colliding colonies) and show that shear forces acting between cells can significantly lower the rate of plasmid transfer. My results can explain why conjugation occurs very rarely in some of these scenarios investigates in laboratory assays.

La première partie de ma thèse est consacrée à l’étude de l’auto-organisation de souches génétiquement modifiées de bactéries Escherichia coli. Ce projet, réalisé en collaboration avec des biologistes synthétiques de l’Université de Hong Kong, a pour objectif l’exploration et le décryptage d’un nouveau mécanisme d’auto-organisation dans des colonies bactériennes multi-espèces. Cela a été inspiré par la question fascinante de comment les écosystèmes bactériens comprenant plusieurs espèces de bactéries peuvent s’auto-organiser dans l’espace. En considérant des systèmes dans lesquels deux souches de bactéries régulent mutuellement leurs motilités, j’ai pu montrer que le contrôle de densité réciproque est une voie générique de formation de motifs: si deux souches tendent à faire augmenter mutuellement leur motilité (la souche A se déplace plus vite quand la souche B est présent, et vice versa), ils subissent un processus de formation de motifs conduisant à la démixtion entre les deux souches. Inversement, l’inhibition mutuelle de la motilité conduit à la formation de motifs avec colocalisation. Ces résultats ont étévalidés expérimentalement par nos collaborateurs biologistes. Par la suite, j’ai étendu mon étude à des systèmes composés de plus de deux espèces en interaction, trouvant des règles simples permettant de prédire l’auto-organisation spatiale d’un nombre arbitraire d’espèces dont la motilité est sous contrôle mutuel. Cette partie de ma thèse ouvre une nouvelle voie pour comprendre l’auto-organisation des colonies bactériennes avec des souches concurrentes, ce qui est une question importante pour comprendre la dynamique des biofilms ou des écosystèmes bactériens dans les sols. Le deuxième problème traité dans ma thèse est inspiré par le comportement collectif des moteurs moléculaires se déplaçant le long des microtubules dans le cytoplasme des cellules eucaryotes. Un modèle pertinent pour le mouvement des moteurs moléculaires est donné par un système paradigmatique de non-équilibre appelé Processus Asymmetrique d’Exclusion Simple, en anglais Asymmetric Simple Exclusion Process (ASEP). Dans ce modèle sur réseau unidimensionnel, les particules se déplacent dans les sites voisins vides à des taux constants, avec un biais gauche-droite qui déséquilibre le système.Lorsqu’il est connecté à ses extrémités à des réservoirs de particules, l’ASEP est un exemple prototypique de transitions de phase unidimensionnelles guidées par les conditions aux limites. Les exemples réalistes, cependant, impliquent rarement une seule voie:les microtubules sont constitués de plusieurs pistes de tubuline auxquelles les moteurs peuvent s’attacher. Dans ma thèse, j’explique comment on peut théoriquement prédire le comportement de phase de systèmes à plusieurs voies complexes, dans lesquels les particules peuvent également sauter entre des voies parallèles. En particulier, je montre que la transition de phase unidimensionnelle vue dans l’ASEP survit cette complexité supplémentaire mais implique de nouvelles caractéristiques telles que des courants transversaux stables non-nulles et une localisation de cisaillement. / The first part of my thesis is devoted to studying the self-organization of engineered strains of run-and-tumble bacteria Escherichia coli. This project, carried out in collaboration with synthetic biologists at Hong Kong University, has as its objective the exploration and decipherment of a novel self-organization mechanism in multi-species bacterial colonies. This was inspired by the fascinating question of how bacterial ecosystems comprising several species of bacteria can self-organize in space. By considering systems in which two strains of bacteria mutually regulate their motilities, I was able to show that reciprocal density control is a generic pattern-formation pathway: if two strains tend tomutually enhance their motility (strain A moves faster when strain B is present, and conversely),they undergo a pattern formation process leading to demixing between the two strains. Conversely, mutual inhibition of motility leads to pattern formation with colocalization. These results were validated experimentally by our biologist collaborators. Subsequently, I extended my study to systems composed of more than two interacting species, finding simple rules that can predict the spatial self-organization of an arbitrary number of species whose motility is under mutual control. This part of my thesis opens up a new route to understand the self-organization of bacterial colonies with competing strains, which is an important question to understand the dynamics of biofilms or bacterial ecosystems in soils.The second problem treated in my thesis is inspired by the collective behaviour ofmolecular motorsmoving along microtubules in the cytoplasm of eukaryotic cells. A relevant model for the molecularmotors’ motion is given by a paradigmatic non-equilibrium system called Asymmetric Simple Exclusion Process (ASEP). In this one-dimensional lattice- based model, particles hop on empty neighboring sites at constant rates, with a leftright bias that drives the systemout of equilibrium. When connected at its ends to particle reservoirs, the ASEP is a prototypical example of one-dimensional boundary driven phase transitions. Realistic examples, however, seldom involve only one lane: microtubules are made of several tubulin tracks to which the motors can attach. In my thesis, I explained how one can theoretically predict the phase behaviour of complex multilane systems, in which particles can also hop between parallel lanes. In particular, I showed that the onedimensional phase transition seen in the ASEP survives this additional complexity but involves new features such as non-zero steady transverse currents and shear localization.