For many years, the temporal program governing DNA replication in budding yeast (S. cerevisiae) has been thought to be actively controlled by a deterministic timing mechanism. We present a reanalysis of microarray studies of DNA replication that supports an interpretation where stochastic effects play a key role and where timing is regulated indirectly by stochastic mechanisms, with no active control being required. Using a quantitative modeling approach, we show that it is possible for replication to be highly stochastic at the level of a single replication origin while simultaneously being highly reproducible at the level of population averages seen in microarray studies. We also discuss how one can extract useful information about replication origins from experiments performed on asynchronously replicating populations of cells, a result that opens up a much larger group of experiments to quantitative analysis.

Les algues vertes (Chlorophyte) forment un groupe d'organismes eukaryotes photosynthétiques le plus souvent unicellulaires au sein duquel ont émergé les plantes terrestres. Depuis le séquençage initial de Chlamydomonas reinhardtii terminé en 2007, la génomique des algues vertes a fortement accéléré puisque à l'heure actuelle, pas moins 10 espèces de Chlorophyte ont leur génome entièrement séquencé. Notre équipe a activement participé à trois de ces projets. Je reviendrai donc ici sur nos trois dernieres années d'immersion dans le monde des algues vertes et de leurs génomes en présentant des données comparatives qui révèlent une grande variabilité dans leurs métabolismes et l'organisation de l'information génétique, bien loin de ce que laissait présager naïvement leur morphologie de "simple petit rond vert". Les algues vertes sont aussi infectés par de grands virus à ADN double brin, les phycoDNAvirus. Ces derniers jouent probablement un très grand rôle dans le contrôle des populations algales. Je présenterai quelques résultats…

Unraveling biological networks from experimental data and prior knowledge can be expressed in the framework of statistical learning. We consider two kinds of networks : gene regulatory networks which are considered as dynamical systems, and protein-protein network which are not a biological object per se but a convenient representation of a set of possible physical relations between proteins. Targetting network inference in both cases, we present two approaches we have developed for a few years: modeling approaches and predictive approaches. Modeling approaches are based on a dynamical model of the behaviour of gene regulatory networks as dynamical systems. We have chosen the probabilistic framework of state-space models or hidden Markov models to represent the evolution of gene regulatory networks and we show how frequentist and Bayesian approches provide very good estimation of dynamics parameters. Results are provided on experimental data ( on microorganisms) as well as toy data such as DREAM data…

Many aspects of genome evolution are best captured by numerical quantities. Examples include genome size, eukaryotic intron density, and so-called phylogenetic profiles (number of genes within the same homolog family across related organisms). I will describe computational methods tailored for the analysis of such numerical characters evolving along a known evolutionary tree. The presentation focuses on inferring general evolutionary trends, as well as information about ancestral states. I will discuss probabilistic frameworks applied to the evolutionary analysis of gene repertoire and exon-intron architecture. Analyses of intron evolution point to intron-rich eukaryotic ancestors. Reconstructions of genome-wide gene repertoires among Archaea reveal the persistence of streamlining processes, punctuated with occasional surges of gene content that (presumably) demarcate episodes of rapid adaptations and population bottlenecks. Perhaps surprisingly, both cases thus illustrate the formative influence of loss in the history of contemporary genomes.

The advances in experimental techniques are allowing the collection of large amount of biological data. The accurate analysis of the available information is important to model biological processes and reduce the costs of the experiments, performing selected computationally driven tests. Two interesting examples of this trend are the large amount of human genetic variation data from high-throughput sequencing experiments and the increasing number of macromolecular structures known at atomic level deposited in the Protein Data Bank. In this talk, I will summarize the results of my research activity during the last years. First, I will present new algorithms for the prediction of effects of single amino acid polymorphisms (SAPs), describing different methods for the prediction of the protein stability change upon single point mutation [1] and for the prediction of their effects on human health [2,3] using protein sequence and structural information. In the last part of the talk I will present…

Many proteins recognize specific sequence motifs in DNA and form stable protein-DNA complexes to perform a range of biological functions associated with transcription, repair and packaging. Although, DNA sequences recognized by most proteins are relatively easily described by small sequence motifs, similar motif-like alphabet in proteins is rarely observed. Sequence and structural elements of proteins interact with DNA in a complex, cooperative and dynamic manner. We have been investigating sequence, structural, evolutionary and thermodynamic patterns of DNA-binding proteins to establish the broad principles of DNA-recognition by them. Our studies have revealed several interesting features useful for understanding and predicting DNA-binding proteins, their binding sites and pinpoint the most significant players in the stability of these complexes.We have developed these leads into novel algorithms and web-based tools for public use.

.