The genomic diversity of viral quasispecies is a subject of great interest, especially for chronic infections. Characterization of viral diversity can be addressed by high-throughput sequencing technology (454 Life Sciences, Illumina, SOLiD, Ion Torrent, etc.). Standard assembly software was originally designed for single genome assembly and cannot be used to assemble and estimate the frequency of closely related quasispecies sequences. This work focuses on parsimonious and maximum likelihood models for assembling viral quasispecies and estimating their frequencies from 454 sequencing data. Our methods have been applied to several RNA viruses (HCV, IBV) as well as DNA viruses (HBV), genotyped using 454 Life Sciences amplicon and shotgun methods.

The genomic diversity of viral quasispecies is a subject of great interest, especially for chronic infections. Characterization of viral diversity can be addressed by high-throughput sequencing technology (454 Life Sciences, Illumina, SOLiD, Ion Torrent, etc.). Standard assembly software was originally designed for single genome assembly and cannot be used to assemble and estimate the frequency of closely related quasispecies sequences. This work focuses on parsimonious and maximum likelihood models for assembling viral quasispecies and estimating their frequencies from 454 sequencing data. Our methods have been applied to several RNA viruses (HCV, IBV) as well as DNA viruses (HBV), genotyped using 454 Life Sciences amplicon and shotgun methods.

Recent advances in next generation sequencing have provided unprecedented opportunities for high-throughput genomic research, inexpensively producing millions of genomic sequences in a single run. Analysis of massive volumes of data results in a more accurate picture of the genome complexity and requires adequate bioinformatics support. We explore computational challenges of applying next generation sequencing to particular applications, focusing on the problem of reconstructing viral quasispecies spectrum from pyrosequencing shotgun reads and problem of inferring informative single nucleotide polymorphisms (SNPs), statistically covering genetic variation of a genome region in genome-wide association studies. The genomic diversity of viral quasispecies is a subject of a great interest, particularly for chronic infections, since it can lead to resistance to existing therapies. High-throughput sequencing is a promising approach to characterizing viral diversity, but unfortunately standard assembly software cannot be used to simultaneously assemble and estimate the abundance of multiple closely related (but non-identical) quasispecies sequences. Here, we introduce a new Viral Spectrum Assembler (ViSpA) for inferring quasispecies spectrum and compare it with the state-of-the-art ShoRAH tool on both synthetic and real 454 pyrosequencing shotgun reads from HCV and HIV quasispecies. While ShoRAH has an advanced error correction algorithm, ViSpA is better at quasispecies assembling, producing more accurate reconstruction of a viral population. We also foresee ViSpA application to the analysis of high-throughput sequencing data from bacterial metagenomic samples and ecological samples of eukaryote populations. Due to the large data volume in genome-wide association studies, it is desirable to find a small subset of SNPs (tags) that covers the genetic variation of the entire set. We explore the trade-off between the number of tags used per non-tagged SNP and possible overfitting and propose an efficient 2LR-Tagging heuristic.

Les arbovirus (dont le virus chikungunya, CHIKV) sont responsables de millions d'infections chaque année ; aucun vaccin n'est encore approuvé, et les traitements disponibles restent limités. De part leur circulation constante entre le moustique et l'humain, leur adaptation rapide à différents hôtes est un facteur clé pour leur pathogenèse. Le taux d'erreur particulièrement élevé de leur polymérase ARN permet une rapide diversification génique qui conduit à la génération d'un nuages de mutants, appelée quasi espèce. Les quasi espèces contiennent non seulement des génomes mutés, mais aussi des ARN recombinés à partir de deux génomes d'origine différente, ainsi que des génomes avec de grandes délétions, incapables de se répliquer sans l'aide d'un autre virion qui doit infection la même cellule, nommés génomes viraux défectifs (GVD). Une régulation étroite de la taille du nuage de mutants est clé pour une pathogenèse efficace : si trop petit, le potentiel adaptatif du virus sera impacté ; au contraire, une quasi-espèce trop grande peut mener à l'accumulation rapide de mutations délétères pour le virus. Alors que la régulation du paysage mutationnel est atteinte grâce à un taux d'erreur de la polymérase viral finement contrôlé, la recombinaison et la réplication des génomes défectifs sont influencés par le potentiel de co-infection des cellules cibles. Dans ce contexte, l'exclusion de la surinfection (ESI), un processus par lequel l'infection par un premier virus inhibe l'infection par un second virus, peut influer le dynamique de la quasi-espèce. Bien que décrite dans la plupart des familles virales, les mécanismes à l'origine de l'ESI restent mal caractérisés. Dans ce travail, je montre que CHIKV exclut une infection future par CHIKV, mais aussi par le virus Sindbis et le virus de la grippe A, mais non par le virus du Nil occidental. Je démontre que l'exclusion de CHIKV se situe au niveau de la pénétration du génome viral dans le cytoplasme, puis de sa réplication, mais n'influence ni l'attachement du virion ni la traduction de son génome. Je montre également que l'ESI est indépendant de l'action des interférons de type I, et qu'elle n'est médiée ni par la transcription cellulaire, ni par un facteur soluble. De plus, l'exclusion n'est pas due à une unique protéine virale, suggérant un potentiel rôle de la réponse cellulaire à l'infection.Déterminer l'influence des pressions immunologiques dans l'établissement de la quasi-espèce peut aider à une meilleure compréhension de l'interaction entre évolution virale et réponse immunitaire. Bien que la caractérisation non biaisée des mutations ponctuelles fût le fruit de nombreux travaux, les GVD restent peu caractérisées, en particulier chez les alphavirus. Dans la deuxième partie de ce travail, je développe des outils bio-informatiques pour isoler rapidement les GVD de données de séquençage à haut débit, et analyse les avantages et les inconvénients d'un ajout d'une étape de pré-amplification pour détecter et quantifier les GVD. À l'aide de ces outils, je fournis ensuite la première description complète des GVD produits par des passages séquentiels de CHIKV en culture cellulaire. En particulier, je montre que le type de GVD générés est très dépendants du type cellulaire, avec des motifs de séquences et des cadres de lecture ouverts différents lorsque les cellules hôtes sont des cellules de mammifère ou d'insecte. Ces résultats soulignent le role de l'environnement cellulaire dans le modelage de la quasi-espèce, et des GVD en particulier. Des travaux futurs aideront à dévoiler les mécanismes sous-jacents à cette interaction et pourraient permettre la conception de nouvelles stratégies thérapeutiques ciblant les dynamiques des quasi-espèces. / Arboviruses such as chikungunya virus (CHIKV) are responsible for millions of yearly infections, with no approved vaccines and limited treatments. Because they circulate between mosquitoes and humans, their fast adaptation potential to different hosts is key to pathogenesis. To achieve genome diversification, they rely on the error-prone nature of their self-encoded RNA-dependent RNA polymerase, which quickly generates a cloud of mutants, termed quasispecies. Quasispecies contain not only mutated genomes, but also shuffled genomes of different parental origin (through a process known as recombination), as well as genomes with large deletions, unable to replicate without the co-infection with a full-length helper genome, and thus termed defective viral genomes (DVGs). A tight regulation of the mutant cloud size is key to pathogenesis: if too small, it will limit the adaptation potential of the virus, whilst too big a quasispecies may lead to the fast accumulation of deleterious mutations. While regulation of the mutational landscape is achieved through the finely tuned error rate of the viral polymerase, recombination and DVG replication are influenced by the co-infection potential of the target cells.In this context, superinfection exclusion (SIE), a process by which infection by a first virus prevents infection by a second, closely related virus, can regulate quasispecies dynamics. While described in most viral families, mechanisms underlying SIE remain poorly characterised. Here, I show that CHIKV infection excludes subsequent infection by CHIKV, Sindbis virus and influenza A virus, but not West Nile virus. I demonstrate that CHIKV exclusion occurs at two steps, impacting independently viral penetration and replication, but does not directly influence binding, nor viral protein translation. I further show that SIE is interferon independent, and does not rely on host cell transcription nor on soluble cellular factors. Moreover, exclusion is not mediated by the action of a single CHIKV protein, suggesting that a cellular response may be at play. Assessing how different immunological pressures can shape quasispecies landscape may prove useful to a more thorough understanding of the interplay between viral evolution and the immune response. Although the unbiased study of point mutations has received much attention, less is known about the characteristics of DVGs, especially in alphaviruses. In the second part of this work, I develop bioinformatics tools to quickly isolate DVGs from next-generation sequencing data, and assess the advantages and drawbacks of pre-amplification steps to detect and quantify DVGs. Using these tools, I provide the first unbiased description of the DVG landscape generated through serial passaging of CHIKV in cell culture. In particular, I show that the DVG landscape is highly dependent on the cell type, with sequence patterns and open reading frames differing between DVGs generated in mammalian and insect cells. These results highlight the role of the cellular environment in shaping quasispecies, and DVGs in particular. Future work will help uncover the mechanisms underlying this crosstalk and may prove useful for the design of treatments targeting quasispecies dynamics.