No description available.

In this thesis, a highly comparative framework for time-series analysis is developed. The approach draws on large, interdisciplinary collections of over 9000 time-series analysis methods, or operations, and over 30 000 time series, which we have assembled. Statistical learning methods were used to analyze structure in the set of operations applied to the time series, allowing us to relate different types of scientific methods to one another, and to investigate redundancy across them. An analogous process applied to the data allowed different types of time series to be linked based on their properties, and in particular to connect time series generated by theoretical models with those measured from relevant real-world systems. In the remainder of the thesis, methods for addressing specific problems in time-series analysis are presented that use our diverse collection of operations to represent time series in terms of their measured properties. The broad utility of this highly comparative approach is demonstrated using various case studies, including the discrimination of pathological heart beat series, classification of Parkinsonian phonemes, estimation of the scaling exponent of self-affine time series, prediction of cord pH from fetal heart rates recorded during labor, and the assignment of emotional content to speech recordings. Our methods are also applied to labeled datasets of short time-series patterns studied in temporal data mining, where our feature-based approach exhibits benefits over conventional time-domain classifiers. Lastly, a feature-based dimensionality reduction framework is developed that links dependencies measured between operations to the number of free parameters in a time-series model that could be used to generate a time-series dataset.

No description available.

Un défi de la transcriptomique est d'explorer l'intégralité du répertoire des transcrits normaux et anormaux. Les analyses de GEP (Gene Expression Profiling) basées sur la technologie des puces à ADN sont largement exploitées en cancérologie depuis plusieurs années. Parallèlement, les nouvelles méthodes basées sur le séquençage à haut débit offrent désormais la possibilité de réaliser des analyses précises et sensibles nécessaires à l'étude des cellules normales et cancéreuses. Quelle que soit la méthode, la caractérisation de l'ensemble des transcrits codants et non-codants représente un réel défi biologique pour la recherche de nouveaux marqueurs de diagnostic et de pronostic, et pour la bonne prise en charge des patients. Dans ce travail, j'ai eu l'occasion de traiter deux aspects différents de la biologie qui convergent vers l'identification de transcrits exprimés de manière aberrante dans les leucémies myéloïdes. Le premier aspect a consisté à proposer une sélection de biomarqueurs moléculaires pour la caractérisation de la leucémie myélomonocytaire chronique (LMMC). A partir de l'expression de ces gènes, nous avons développé un score de pronostic qui a permis de définir deux groupes de patients cliniquement distincts. Nous avons ensuite complété notre étude par une caractérisation phénotypique par cytométrie en flux des sous-populations cellulaires aberrantes constituant les lignages mono- et granulocytaires. Une partie de ce travail a été étendue aux leucémies aiguës myéloïdes (LAM) à caryotype normal (CN). L'autre aspect a consisté à participer à la mise en place d'une approche computationnelle intégrée pour caractériser de nouveaux ARNs non annotés et fort probablement non-codants. En explorant des données de Digital Gene Expression (DGE), nous avons quantifié et caractérisé la fraction de ces transcrits dans les régions intergéniques. Nous avons vérifié l'expression de ces nouveaux transcrits dans les tissus normaux et cancéreux en croisant avec d'autres données d'expression, telles que le RNA-Sequencing, et regarder leur conservation et leur expression dans d'autres espèces. / A challenge of transcriptomics is to explore the full repertoire of normal and abnormal transcripts. Gene expression profiling analyses based on microarray technology are widely used in cancer research since many years. Meanwhile, new methods based on high-throughput sequencing methods offers henceforth the possibility to undergo accurate and sensitive analyses necessary for studying normal and cancer cells. Despite the method, the characterization of all coding and non-coding transcripts is a real biological challenge in identifying novel diagnostic and prognostic markers, and for the proper care of patients. In the present work, I had the opportunity to address two different aspects of biology, both convergent toward the identification of aberrantly expressed transcripts in myeloid leukemia. The first aspect was to provide a selection of molecular biomarkers for the characterization of chronic myelomonocytic leukemia (CMML). We developed a gene expression-based prognostic score which identified two clinically distinct groups of patients. We then completed our study with a phenotypic characterization by flow cytometry of aberrant subpopulations constituting the myeloid and granulocytic lineages. A part of this work has been extended to acute myeloid leukemia (AML) patients with normal karyotype. The other aspect was to participate in the implementation of an integrated computational approach in order to characterize novel non annotated RNAs, more likely non-coding. We quantified and characterized the proportion of these transcripts in intergenic regions by exploring Digital Gene Expression (DGE) data. We checked their expression in normal and cancer tissues by crossing with RNA-Seq data, and their conservation and expression in other species.