Search

Constraints on practice can benefit motor learning by guiding the learner towards efficient coordination patterns, but can also narrow the potential solution space of coordination and control. The aim of this paper was to investigate whether narrowing the solution space through more restrictive task constraints limits the expression of potential exploratory behaviours during the learning process, identified using Drifting Markov Models. The change in interlimb coordination of 7 learners practicing for 16 lessons over 2 months was analysed to quantify motor exploration and identify periods of metastable regimes of coordination. Results showed that the observed exploratory dynamics were highly individual both in terms of range of exploration and in the patterns of search. The more restrictive task constraints did not impair the range of exploration but rather channelled the exploration around a few selected patterns. In addition, restraining the nature of the exploratory process increased the inter-individual differences of the learning rate. Although manipulating the task constraints during learning can help learners to escape from the behavioural consequences of their intrinsic dynamics, maintaining a broad solution space for a diversity of coordination patterns to emerge was key to fostering effective exploration of individual coordination solutions.

International audience; microRNAs are noncoding RNAs which downregulate a large number of target mRNAs and modulate cell activity. Despite continued progress, bioinformatics prediction of microRNA targets remains a challenge since available software still suffer from a lack of accuracy and sensitivity. Moreover, these tools show fairly inconsistent results from one another. Thus, in an attempt to circumvent these difficulties, we aggregated all human results of four important prediction algorithms (miRanda, PITA, SVmicrO, and TargetScan) showing additional characteristics in order to rerank them into a single list. Instead of deciding which prediction tool to use, our method clearly helps biologists getting the best microRNA target predictions from all aggregated databases. The resulting database is freely available through a webtool called miRabel which can take either a list of miRNAs, genes, or signaling pathways as search inputs. Receiver operating characteristic curves and precision-recall curves analysis carried out using experimentally validated data and very large data sets show that miRabel significantly improves the prediction of miRNA targets compared to the four algorithms used separately. Moreover, using the same analytical methods, miRabel shows significantly better predictions than other popular algorithms such as MBSTAR, miRWalk, ExprTarget and miRMap. Interestingly, an F-score analysis revealed that miRabel also significantly improves the relevance of the top results. The aggregation of results from different databases is therefore a powerful and generalizable approach to many other species to improve miRNA target predictions. Thus, miRabel is an efficient tool to guide biologists in their search for miRNA targets and integrate them into a biological context.

International audience; Phaeodactylum tricornutum is the most studied diatom encountered principally in coastal unstable environments. It has been hypothesized that the great adaptability of P. tricornutum is probably due to its pleomorphism. Indeed, P. tricornutum is an atypical diatom since it can display three morphotypes: fusiform, triradiate and oval. Currently, little information is available regarding the physiological significance of this morphogenesis. In this study, we adapted P. tricornutum Pt3 strain to obtain algal culture particularly enriched in one dominant morphotype: fusiform, triradiate or oval. These cultures were used to run high-throughput RNA-Sequencing. The whole mRNA transcriptome of each morphotype was determined. Pairwise comparisons highlighted biological processes and molecular functions which are up-and down-regulated. Finally, intersection analysis allowed us to identify the specific features from the oval morphotype which is of particular interest as it is often described to be more resistant to stresses. This study represent the first transcriptome wide characterization of the three morphotypes from P. tricornutum performed on cultures specifically enriched issued from the same Pt3 strain. This work represents an important step for the understanding of the morphogenesis in P. tricornutum and highlights the particular features of the oval morphotype. Diatoms are a major component of phytoplankton communities. They belong to a group of unicellular heterokont microalgae expected to include at least 200,000 species 1. Their genomes contain unique combinations of nutrient assimilation and metabolic pathways which have been attributed to their origin 2. Indeed, diatoms derived from a serial secondary endosymbiotic event in which a green algae and subsequently a red one were engulfed by a heter-otrophic eukaryote 3-5. Among the diatom species, Thalassiosira pseudonana (T. pseudonana) and Phaeodactylum tricornutum (P. tricornutum; Pt) have been chosen based on the number of research groups working on them and genome size considerations 6-8. The comparative analysis of T. pseudonana and P. tricornutum genomes highlights the presence of 10% of diatom specific genes which have no homologs in other eukaryotes 6,7,9. In addition, it has been demonstrated that P. tricornutum and T. pseudonana possess 6.4% and 2% of transposable elements, respectively 6,7,10 as well as microRNAs 11,12. Moreover, a recent re-analysis of the P. tricornutum genome revealed that among 251 different patterns identified across the entire genome, many genes have limited evolutionary conservation with 47% having originated within the recent vertical history of the stramenopile lineage and a total of 26% of genes (3170 genes) being specific to P. tricornutum 8. P. tricornutum is probably the widest studied diatom. Its genome has been sequenced and a large scale tran-scriptomic analysis has been recently published 7-9. It is a cosmopolitan diatom not widely distributed in nature, but is encountered principally in coastal unstable environments such as estuaries or rock pools 13,14. Moreover, P.

International audience; Semi-Markov models, independently introduced by Lévy (1954), Smith (1955) and Takacs (1954), are a generalization of the well-known Markov models. For semi-Markov models, sojourn times can be arbitrarily distributed, while sojourn times of Markov models are constrained to be exponentially distributed (in continuous time) or geometrically distributed (in discrete time). The aim of this paper is to present the R package SMM, devoted to the simulation and estimation of discrete-time multi-state semi-Markov and Markov models. For the semi-Markov case we have considered: parametric and non-parametric estimation; with and without censoring at the beginning and/or at the end of sample paths; one or several independent sample paths. Several discrete-time distributions are considered for the parametric estimation of sojourn time distributions of semi-Markov chains: Uniform, Geometric, Poisson, Discrete Weibull and Binomial Negative.

In this work we focus on multi-state systems modeled by means of a particular class of non-homogeneous Markov processes introduced in Vergne (Stat Appl Genet Mol Biol 7(1):1–45, 2008), called drifting Markov processes. The main idea behind this type of processes is to consider a non-homogeneity that is “smooth”, of a known shape. More precisely, the Markov transition matrix is assumed to be a linear (polynomial) function of two (several) Markov transition matrices. For this class of systems, we first obtain explicit expressions for reliability/survival indicators of drifting Markov models, like reliability, availability, maintainability and failure rates. Then, under different statistical settings, we estimate the parameters of the model, obtain plug-in estimators of the associated reliability/survival indicators and investigate the consistency of the estimators. The quality of the proposed estimators and the model validation is illustrated through numerical experiments.

Background In the current context of personalized medicine, one of the major challenges in the management of rheumatoid arthritis (RA) is to identify biomarkers that predict drug responsiveness. From the European APPRAISE trial, our main objective was to identify a gene expression profile associated with responsiveness to abatacept (ABA) + methotrexate (MTX) and to understand the involvement of this signature in the pathophysiology of RA. Methods Whole human genome microarrays (4 × 44 K) were performed from a first subset of 36 patients with RA. Data validation by quantitative reverse-transcription (qRT)-PCR was performed from a second independent subset of 32 patients with RA. Gene Ontology and WikiPathways database allowed us to highlight the specific biological mechanisms involved in predicting response to ABA/MTX. Results From the first subset of 36 patients with RA, a combination including 87 transcripts allowed almost perfect separation between responders and non-responders to ABA/MTX. Next, the second subset of patients 32 with RA allowed validation by qRT-PCR of a minimal signature with only four genes. This latter signature categorized 81% of patients with RA with 75% sensitivity, 85% specificity and 85% negative predictive value. This combination showed a significant enrichment of genes involved in electron transport chain (ETC) pathways. Seven transcripts from ETC pathways (NDUFA6, NDUFA4, UQCRQ, ATP5J, COX7A2, COX7B, COX6A1) were significantly downregulated in responders versus non-responders to ABA/MTX. Moreover, dysregulation of these genes was independent of inflammation and was specific to ABA response. Conclusion Pre-silencing of ETC genes is associated with future response to ABA/MTX and might be a crucial key to susceptibility to ABA response. Electronic supplementary material The online version of this article (doi:10.1186/s13075-017-1319-8) contains supplementary material, which is available to authorized users.

DNA barcoding is the assignment of individuals to species using standardized mitochondrial sequences. Nuclear data are sometimes added to the mitochondrial data to increase power. A barcoding method for analysing mitochondrial and nuclear data is developed. It is a Bayesian method based on the coalescent model. Then this method is assessed using simulated and real data. It is found that adding nuclear data can reduce the number of ambiguous assignments. Finally, the robustness of coalescent-based barcoding to departures from model assumptions is studied using simulations. This method is found to be robust to past population size variations, to within-species population structures, and to designs that poorly sample populations within species. Supplementary Material is available online at www.liebertonline.com/cmb.

We describe the statistics of repetition times of a string of symbols in a stochastic process. Denote by T(A) the time elapsed until the process spells the finite string A and by S(A) the number of consecutive repetitions of A. We prove that, if the length of the string grows unbondedly, (1) the distribution of T(A), when the process starts with A, is well aproximated by a certain mixture of the point measure at the origin and an exponential law, and (2) S(A) is approximately geometrically distributed. We provide sharp error terms for each of these approximations. The errors we obtain are point-wise and allow to get also approximations for all the moments of T(A) and S(A). To obtain (1) we assume that the process is phi-mixing while to obtain (2) we assume the convergence of certain contidional probabilities.

International audience; We describe the statistics of the number of occurrences of a string of symbols in a stochastic process: taking a string A of length n, we prove that the number of visits to A up to time t, denoted by N t , has approximately a Poisson distribution. We provide a sharp error for this approximation. In contrast to previous works which present uniform error terms based on the total variation distance, our error is point-wise. As a byproduct we obtain that all the moments of N t are finite. Moreover, we obtain explicit approximations for all of them. Our result holds for processes that verify the φ-mixing condition. The error term is explicitly expressed as a function of the rate function φ and is easily computable.

In this article, we introduce the drifting Markov models (DMMs) which are inhomogeneous Markov models designed for modeling the heterogeneities of sequences (in our case DNA or protein sequences) in a more flexible way than homogeneous Markov chains or even hidden Markov models (HMMs). We focus here on the polynomial drift: the transition matrix varies in a polynomial way. To show the reliability of our models on DNA, we exhibit high similarities between the probability distributions of nucleotides obtained by our models and the frequencies of these nucleotides computed by using a sliding window. In a further step, these DMMs can be used as the states of an HMM: on each of its segments, the observed process can be modeled by a drifting Markov model. Search of rare words in DNA sequences remains possible with DMMs and according to the fits provided, DMMs turn out to be a powerful tool for this purpose. The software is available on request from the author. It will soon be integrated on seq++ library (http://stat.genopole.cnrs.fr/seqpp/).

En finance, la calibration d’un modèle financier joue un rôle important pour la gestion des risques et pour les produits dérivés, et nécessite d’être effectuée rapidement et avec précision. Un algorithme d’optimisation utilisé avec un réseau de neurones artificiels (RNA) permet de réaliser cet objectif. Pour ce faire, j’ai d’abord choisi une méthode classique de pricing pour générer des données d’apprentissage, puis entraîné un RNA pour remplacer la méthode de pricing. Le principe du RNA est aussi présenté dans ce rapport. La calibration s’est faite à l’aide de l’algorithme d’évolution différentielle, qui cherche le minimum global d’erreur entre la volatilité implicite (VI) prédite par le RNA et la VI observée sur le marché financier. Mon travail a permis à l’équipe de statistique du LMRS d’avoir un répertoire des méthodes de pricing et de calibration, de s’initier au machine learning (plus particulièrement au deep learning) et d’apprendre la construction des RNA. Les résultats motivent une étude plus approfondie des méthodes d’apprentissage du RNA afin d’améliorer les performances en terme de temps de calcul et de précision.