University of Grenoble

About: University of Grenoble is a education organization based out in Saint-Martin-d'Hères, France. It is known for research contribution in the topics: Population & Context (language use). The organization has 25658 authors who have published 45143 publications receiving 909760 citations.

Flexible-meccano

Abstract: Motivation: Intrinsically disordered proteins (IDPs) represent a significant fraction of the human proteome. The classical structure function paradigm that has successfully underpinned our understanding of molecular biology breaks down when considering proteins that have no stable tertiary structure in their functional form. One convenient approach is to describe the protein in terms of an equilibrium of rapidly inter-converting conformers. Currently, tools to generate such ensemble descriptions are extremely rare, and poorly adapted to the prediction of experimental data. Results: We present flexible-meccano—a highly efficient algorithm that generates ensembles of molecules, on the basis of amino acid-specific conformational potentials and volume exclusion. Conformational sampling depends uniquely on the primary sequence, with the possibility of introducing additional local or long-range conformational propensities at an amino acid-specific resolution. The algorithm can also be used to calculate expected values of experimental parameters measured at atomic or molecular resolution, such as nuclear magnetic resonance (NMR) and small angle scattering, respectively. We envisage that flexible-meccano will be useful for researchers who wish to compare experimental data with those expected from a fully disordered protein, researchers who see experimental evidence of deviation from ‘random coil’ behaviour in their protein, or researchers who are interested in working with a broad ensemble of conformers representing the flexibility of the IDP of interest. Availability: A fully documented multi-platform executable is provided, with examples, at http://www.ibs.fr/science-213/scientific-output/software/flexible-meccano/ Contact: martin.blackledge@ibs.fr

How Good Is My GAN

Abstract: Generative adversarial networks (GANs) are one of the most popular methods for generating images today. While impressive results have been validated by visual inspection, a number of quantitative criteria have emerged only recently. We argue here that the existing ones are insufficient and need to be in adequation with the task at hand. In this paper we introduce two measures based on image classification—GAN-train and GAN-test, which approximate the recall (diversity) and precision (quality of the image) of GANs respectively. We evaluate a number of recent GAN approaches based on these two measures and demonstrate a clear difference in performance. Furthermore, we observe that the increasing difficulty of the dataset, from CIFAR10 over CIFAR100 to ImageNet, shows an inverse correlation with the quality of the GANs, as clearly evident from our measures.

APP, PSEN1, and PSEN2 mutations in early-onset Alzheimer disease: A genetic screening study of familial and sporadic cases

Abstract: Background Amyloid protein precursor (APP), presenilin-1 (PSEN1), and presenilin-2 (PSEN2) mutations cause autosomal dominant forms of early-onset Alzheimer disease (AD-EOAD). Although these genes were identified in the 1990s, variant classification remains a challenge, highlighting the need to colligate mutations from large series. Methods and findings We report here a novel update (2012–2016) of the genetic screening of the large AD-EOAD series ascertained across 28 French hospitals from 1993 onwards, bringing the total number of families with identified mutations to n = 170. Families were included when at least two first-degree relatives suffered from early-onset Alzheimer disease (EOAD) with an age of onset (AOO) ≤65 y in two generations. Furthermore, we also screened 129 sporadic cases of Alzheimer disease with an AOO below age 51 (44% males, mean AOO = 45 ± 2 y). APP, PSEN1, or PSEN2 mutations were identified in 53 novel AD-EOAD families. Of the 129 sporadic cases screened, 17 carried a PSEN1 mutation and 1 carried an APP duplication (13%). Parental DNA was available for 10 sporadic mutation carriers, allowing us to show that the mutation had occurred de novo in each case. Thirteen mutations (12 in PSEN1 and 1 in PSEN2) identified either in familial or in sporadic cases were previously unreported. Of the 53 mutation carriers with available cerebrospinal fluid (CSF) biomarkers, 46 (87%) had all three CSF biomarkers—total tau protein (Tau), phospho-tau protein (P-Tau), and amyloid β (Aβ)42—in abnormal ranges. No mutation carrier had the three biomarkers in normal ranges. One limitation of this study is the absence of functional assessment of the possibly and probably pathogenic variants, which should help their classification. Conclusions Our findings suggest that a nonnegligible fraction of PSEN1 mutations occurs de novo, which is of high importance for genetic counseling, as PSEN1 mutational screening is currently performed in familial cases only. Among the 90 distinct mutations found in the whole sample of families and isolated cases, definite pathogenicity is currently established for only 77%, emphasizing the need to pursue the effort to classify variants.

A priori convergence of the greedy algorithm for the parametrized reduced basis method

Abstract: The convergence and efficiency of the reduced basis method used for the approximation of the solutions to a class of problems written as a parametrized PDE depends heavily on the choice of the elements that constitute the "reduced basis". The purpose of this paper is to analyze the a priori convergence for one of the approaches used for the selection of these elements, the greedy algorithm. Under natural hypothesis on the set of all solutions to the problem obtained when the parameter varies, we prove that three greedy algorithms converge; the last algorithm, based on the use of an a posteriori estimator, is the approach actually employed in the calculations.

Structural model for the biogenic Mn oxide produced by Pseudomonas putida

Abstract: X-ray diffraction (XRD) and Mn K-edge extended X-ray absorption Þ ne structure (EXAFS) spectroscopy were combined to elaborate a structural model for phyllomanganates (layer-type Mn oxides) lacking 3D ordering (turbostratic stacking). These techniques were applied to a sample produced by a common soil and freshwater bacterium (Pseudomonas putida), and to two synthetic analogs, δ-MnO2 and acid birnessite, obtained by the reduction of potassium permanganate with MnCl2 and HCl, respectively. To interpret the diffraction and spectroscopic data, we applied an XRD simulation technique utilized previously for well-crystallized birnessite varieties, complementing this approach with single-scattering-path simulations of the Mn K-edge EXAFS spectra. Our structural analyses revealed that all three Mn oxides have an hexagonal layer symmetry with layers comprising edgesharing Mn 4+ O6 octahedra and cation vacancies, but no layer Mn 3+ O6 octahedra. The proportion of cation vacancies in the layers ranged from 6 to 17%, these vacancies being charge-compensated in the interlayer by protons, alkali metals, and Mn atoms, in amounts that vary with the phyllomanganate species and synthesis medium. Both vacancies and interlayer Mn were most abundant in the biogenic oxide. The diffracting crystallites contained three to six randomly stacked layers and have coherent scattering domains of 19–42 A in the c* direction, and of 60–85 A in the a-b plane. Thus, the Mn oxides investigated here are nanoparticles that bear signiÞ cant permanent structural charge resulting from cation layer vacancies and variable surface charge from unsaturated O atoms at layer edges.