Jean-Luc Giraudel

Gene and Metabolite Regulatory Network Analysis of Early Developing Fruit Tissues Highlights New Candidate Genes for the Control of Tomato Fruit Composition and Development

Variations in early fruit development and composition may have major impacts on the taste and the overall quality of ripe tomato (Solanum lycopersicum) fruit. To get insights into the networks involved in these coordinated processes and to identify key regulatory genes, we explored the transcriptional and metabolic changes in expanding tomato fruit tissues using multivariate analysis and gene-metabolite correlation networks. To this end, we demonstrated and took advantage of the existence of clear structural and compositional differences between expanding mesocarp and locular tissue during fruit development (12–35 d postanthesis). Transcriptome and metabolome analyses were carried out with tomato microarrays and analytical methods including proton nuclear magnetic resonance and liquid chromatography-mass spectrometry, respectively. Pairwise comparisons of metabolite contents and gene expression profiles detected up to 37 direct gene-metabolite correlations involving regulatory genes (e.g. the correlations between glutamine, bZIP, and MYB transcription factors). Correlation network analyses revealed the existence of major hub genes correlated with 10 or more regulatory transcripts and embedded in a large regulatory network. This approach proved to be a valuable strategy for identifying specific subsets of genes implicated in key processes of fruit development and metabolism, which are therefore potential targets for genetic improvement of tomato fruit quality.

Quantitative metabolic profiles of tomato flesh and seeds during fruit development: complementary analysis with ANN and PCA

Tomato, an essential crop in terms of economic importance and nutritional quality, is also used as a model species for all fleshy fruits and for genomics of Solanaceae. Tomato fruit quality at harvest is a direct function of its metabolite content, which in turn is a result of many physiological changes during fruit development. The aim of the work reported here was to develop a global approach to characterize changes in metabolic profiles in two interdependent tissues from the same tomato fruits. Absolute quantification data of compounds in flesh and seeds from 8xa0days to 45xa0days post anthesis (DPA) were obtained through untargeted (proton nuclear magnetic resonance, 1H-NMR) and targeted metabolic profiling (liquid chromatography with diode array detection (LC-DAD) or gas chromatography with flame ionization detection (GC-FID)). These data were analyzed with chemometric approaches. Kohonen self organizing maps (SOM) analysis of these data allowed us to combine multivariate (distribution of samples on Kohonen SOMs) and univariate information (component plane representation of metabolites) in a single analysis. This strategy confirmed published data and brought new insights on tomato flesh and seed composition, thus demonstrating its potential in metabolomics. The compositional changes were related to physiological processes occurring in each tissue. They pointed to (i) some parallel changes at early stages in relation to cell division and transitory storage of carbon, (ii) metabolites participating in the fleshy trait and (iii) metabolites involved in the specific developmental patterns of the seeds.

Predicting stream nitrogen concentration from watershed features using neural networks

The present work describes the development and validation of an artificial neural network (ANN) for the purpose of estimating inorganic and total nitrogen concentrations. The ANN approach has been developed and tested using 927 nonpoint source watersheds studied for relationships between macro-drainage area characteristics and nutrient levels in streams. The ANN had eight independent input variables of watershed parameters (five on land use features, mean annual precipitation, animal unit density and mean stream flow) and two dependent output variables (total and inorganic nitrogen concentrations in the stream). The predictive quality of ANN models was judged with “hold-out” validation procedures. After ANN learning with the training set of data, we obtained a correlation coefficient r of about 0.85 in the testing set. Thus, ANNs are capable of learning the relationships between drainage area characteristics and nitrogen levels in streams, and show a high ability to predict from the new data set. On the basis of the sensitivity analyses we established the relationship between nitrogen concentration and the eight environmental variables.

Spatial distribution and partitioning behavior of selected poly- and perfluoroalkyl substances in freshwater ecosystems: A French nationwide survey

The spatial distribution and partitioning of 22 poly- and perfluoroalkyl substances (PFASs) in 133 selected rivers and lakes were investigated at a nationwide scale in mainland France. ΣPFASs was in the range<LOD-725 ng L(-1) in the dissolved phase (median: 7.9 ng L(-1)) and <LOD-25 ng g(-1) dry weight (dw) in the sediment (median: 0.48 ng g(-1) dw); dissolved PFAS levels were significantly lower at reference sites than at urban, rural or industrial sites. Although perfluorooctane sulfonate (PFOS) was found to be the prevalent compound on average, a multivariate analysis based on neural networks revealed noteworthy trends for other compounds at specific locations and, in some cases, at watershed scale. For instance, several sites along the Rhône River displayed a peculiar PFAS signature, perfluoroalkyl carboxylates (PFCAs) often dominating the PFAS profile (e.g., PFCAs>99% of ΣPFASs in the sediment, likely as a consequence of industrial point source discharge). Several treatments for data below detection limits (non-detects) were used to compute descriptive statistics, differences among groups, and correlations between congeners, as well as log Kd and log Koc partition coefficients; in that respect, the Regression on Order Statistics (robust ROS) method was preferred for descriptive statistics computation while the Akritas-Theil-Sen estimator was used for regression and correlation analyses. Multiple regression results suggest that PFAS levels in the dissolved phase and sediment characteristics (organic carbon fraction and grain size) may be significant controlling factors of PFAS levels in the sediment.

Microsatellites and artificial neural networks: tools for the discrimination between natural and hatchery brown trout (Salmo trutta, L.) in Atlantic populations

Abstract Artificial Neural Networks (ANN) were applied to microsatellite data (highly variable genetic markers) to separate genetically differentiated forms of brown trout ( Salmo trutta ) in south-western France. A classic feed-forward network with one hidden layer was used. Training was performed using a back-propagation algorithm and reference samples representing the different genetic types. The hold-out and the leave-one-out procedures were used to test the validity of the network. They were chosen according to the populations and the questions analysed. The informative content of the different variables used for the distinction (the alleles of the different loci) was also evaluated using the Garson–Goh algorithm. The results of learning gave high percentages of well-classified individuals (up to 95% for the test with the hold-out analysis). This confirms that ANNs are suitable for such genetic analyses of populations. From a biological point of view, the study enabled evaluation of the genetic composition and differentiation of different river populations and of the impact of stocking.

diatSOM: a R-package for diatom biotypology using self-organizing maps

Owing to the high complexity of diatom community data, there is a special need for methods accounting for complex non-linear gradients. A Kohonens self-organizing map (SOM) is a neural network with unsupervised learning. It allows both unbiased classification of the communities and visualization of biological gradients on a two-dimensional plane. However, as with other neural networks, many parameters must be set. A new R-package with a SOM parameterization specifically suited to diatom communities has been developed. Further developments will consist of creating a graphical user interface in order to make this method easier to use for the scientific community.

Artificial neural networks enable discrimination of taxa using morphological characters: an example with a problematic small mammal, the water vole (genus Arvicola) (Rodentia, Cricetidae)

Abstract The aim of this study is to assess the reliability of different artificial neural networks as tools for identifying taxonomic or ecotypic subdivisions in European water voles (genus Arvicola). Self-organizing maps show that taxa at subspecies or species level are organized along morphological gradients correlated with ecotypes (from extreme forms of fossorial Arvicola terrestris to strictly aquatic Arvicola sapidus). An automatic discrimination process based on a multilayer feed-forward network, working with 17 French and one Spanish reference populations, proved able to correctly identify species in test samples from France, but sometimes misidentified specimens from other countries. Three morpho-ecotypes were identified using a self-organizing map with a single-species dataset (Arvicola terrestris only). However, there were no clues as to how these morpho-ecotypes might correlate with the known phylogenetics of the species. Our model can identify species on the basis of a limited range of cranial measurements which are available from vole skull fragments commonly found in owl pellets. This can have very practical risk-assessment applications as it allows easier identification of the more fossorial Arvicola terrestris populations which are a threat to crops.