FORUM: building a Knowledge Graph from public databases and scientific literature to extract associations between chemicals and diseases.

MOTIVATION: Metabolomics studies aim at reporting a metabolic signature (list of metabolites) related to a particular experimental condition. These signatures are instrumental in the identification of biomarkers or classification of individuals, however their biological and physiological interpretation remains a challenge. To support this task, we introduce FORUM: a Knowledge Graph (KG) providing a semantic representation of relations between chemicals and biomedical concepts, built from a federation of life science databases and scientific literature repositories. RESULTS: The use of a Semantic Web framework on biological data allows us to apply ontological-based reasoning to infer new relations between entities. We show that these new relations provide different levels of abstraction and could open the path to new hypotheses. We estimate the statistical relevance of each extracted relation, explicit or inferred, using an enrichment analysis, and instantiate them as new knowledge in the KG to support results interpretation/further inquiries. AVAILABILITY AND IMPLEMENTATION: A web interface to browse and download the extracted relations, as well as a SPARQL endpoint to directly probe the whole FORUM KG, are available at https://forum-webapp.semantic-metabolomics.fr. The code needed to reproduce the triplestore is available at https://github.com/eMetaboHUB/Forum-DiseasesChem. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Cytonuclear interactions remain stable during allopolyploid evolution despite repeated whole-genome duplications in Brassica.

Several plastid macromolecular protein complexes are encoded by both nuclear and plastid genes. Therefore, cytonuclear interactions are held in place to prevent genomic conflicts that may lead to incompatibilities. Allopolyploidy resulting from hybridization and genome doubling of two divergent species can disrupt these fine-tuned interactions, as newly formed allopolyploid species confront biparental nuclear chromosomes with a uniparentally inherited plastid genome. To avoid any deleterious effects of unequal genome inheritance, preferential transcription of the plastid donor over the other donor has been hypothesized to occur in allopolyploids. We used Brassica as a model to study the effects of paleopolyploidy in diploid parental species, as well as the effects of recent and ancient allopolyploidy in Brassica napus, on genes implicated in plastid protein complexes. We first identified redundant nuclear copies involved in those complexes. Compared with cytosolic protein complexes and with genome-wide retention rates, genes involved in plastid protein complexes show a higher retention of genes in duplicated and triplicated copies. Those redundant copies are functional and are undergoing strong purifying selection. We then compared transcription patterns and sequences of those redundant gene copies between resynthesized allopolyploids and their diploid parents. The neopolyploids showed no biased subgenome expression or maternal homogenization via gene conversion, despite the presence of some non-synonymous substitutions between plastid genomes of parental progenitors. Instead, subgenome dominance was observed regardless of the maternal progenitor. Our results provide new insights on the evolution of plastid protein complexes that could be tested and generalized in other allopolyploid species.

QTL detection for coccidiosis (Eimeria tenella) resistance in a Fayoumi × Leghorn F2 cross, using a medium-density SNP panel.

BACKGROUND: Coccidiosis is a major parasitic disease that causes huge economic losses to the poultry industry. Its pathogenicity leads to depression of body weight gain, lesions and, in the most serious cases, death in affected animals. Genetic variability for resistance to coccidiosis in the chicken has been demonstrated and if this natural resistance could be exploited, it would reduce the costs of the disease. Previously, a design to characterize the genetic regulation of Eimeria tenella resistance was set up in a Fayoumi × Leghorn F2 cross. The 860 F2 animals of this design were phenotyped for weight gain, plasma coloration, hematocrit level, intestinal lesion score and body temperature. In the work reported here, the 860 animals were genotyped for a panel of 1393 (157 microsatellites and 1236 single nucleotide polymorphism (SNP) markers that cover the sequenced genome (i.e. the 28 first autosomes and the Z chromosome). In addition, with the aim of finding an index capable of explaining a large amount of the variance associated with resistance to coccidiosis, a composite factor was derived by combining the variables of all these traits in a single variable. QTL detection was performed by linkage analysis using GridQTL and QTLMap. Single and multi-QTL models were applied. RESULTS: Thirty-one QTL were identified i.e. 27 with the single-QTL model and four with the multi-QTL model and the average confidence interval was 5.9 cM. Only a few QTL were common with the previous study that used the same design but focused on the 260 more extreme animals that were genotyped with the 157 microsatellites only. Major differences were also found between results obtained with QTLMap and GridQTL. CONCLUSIONS: The medium-density SNP panel made it possible to genotype new regions of the chicken genome (including micro-chromosomes) that were involved in the genetic control of the traits investigated. This study also highlights the strong variations in QTL detection between different models and marker densities.

Genome-wide interval mapping using SNPs identifies new QTL for growth, body composition and several physiological variables in an F2 intercross between fat and lean chicken lines.

BACKGROUND: For decades, genetic improvement based on measuring growth and body composition traits has been successfully applied in the production of meat-type chickens. However, this conventional approach is hindered by antagonistic genetic correlations between some traits and the high cost of measuring body composition traits. Marker-assisted selection should overcome these problems by selecting loci that have effects on either one trait only or on more than one trait but with a favorable genetic correlation. In the present study, identification of such loci was done by genotyping an F2 intercross between fat and lean lines divergently selected for abdominal fatness genotyped with a medium-density genetic map (120 microsatellites and 1302 single nucleotide polymorphisms). Genome scan linkage analyses were performed for growth (body weight at 1, 3, 5, and 7 weeks, and shank length and diameter at 9 weeks), body composition at 9 weeks (abdominal fat weight and percentage, breast muscle weight and percentage, and thigh weight and percentage), and for several physiological measurements at 7 weeks in the fasting state, i.e. body temperature and plasma levels of IGF-I, NEFA and glucose. Interval mapping analyses were performed with the QTLMap software, including single-trait analyses with single and multiple QTL on the same chromosome. RESULTS: Sixty-seven QTL were detected, most of which had never been described before. Of these 67 QTL, 47 were detected by single-QTL analyses and 20 by multiple-QTL analyses, which underlines the importance of using different statistical models. Close analysis of the genes located in the defined intervals identified several relevant functional candidates, such as ACACA for abdominal fatness, GHSR and GAS1 for breast muscle weight, DCRX and ASPSCR1 for plasma glucose content, and ChEBP for shank diameter. CONCLUSIONS: The medium-density genetic map enabled us to genotype new regions of the chicken genome (including micro-chromosomes) that influenced the traits investigated. With this marker density, confidence intervals were sufficiently small (14 cM on average) to search for candidate genes. Altogether, this new information provides a valuable starting point for the identification of causative genes responsible for important QTL controlling growth, body composition and metabolic traits in the broiler chicken.

Evaluation of GC/MS-Based 13C-Positional Approaches for TMS Derivatives of Organic and Amino Acids and Application to Plant 13C-Labeled Experiments.

Analysis of plant metabolite 13C-enrichments with gas-chromatography mass spectrometry (GC/MS) has gained interest recently. By combining multiple fragments of a trimethylsilyl (TMS) derivative, 13C-positional enrichments can be calculated. However, this new approach may suffer from analytical biases depending on the fragments selected for calculation leading to significant errors in the final results. The goal of this study was to provide a framework for the validation of 13C-positional approaches and their application to plants based on some key metabolites (glycine, serine, glutamate, proline, α-alanine and malate). For this purpose, we used tailor-made 13C-PT standards, harboring known carbon isotopologue distributions and 13C-positional enrichments, to evaluate the reliability of GC-MS measurements and positional calculations. Overall, we showed that some mass fragments of proline_2TMS, glutamate_3TMS, malate_3TMS and α-alanine_2TMS had important biases for 13C measurements resulting in significant errors in the computational estimation of 13C-positional enrichments. Nevertheless, we validated a GC/MS-based 13C-positional approach for the following atomic positions: (i) C1 and C2 of glycine_3TMS, (ii) C1, C2 and C3 of serine_3TMS, and (iii) C1 of malate_3TMS and glutamate_3TMS. We successfully applied this approach to plant 13C-labeled experiments for investigating key metabolic fluxes of plant primary metabolism (photorespiration, tricarboxylic acid cycle and phosphoenolpyruvate carboxylase activity).

Statistical properties of interval mapping methods on quantitative trait loci location: impact on QTL/eQTL analyses.

BACKGROUND: Quantitative trait loci (QTL) detection on a huge amount of phenotypes, like eQTL detection on transcriptomic data, can be dramatically impaired by the statistical properties of interval mapping methods. One of these major outcomes is the high number of QTL detected at marker locations. The present study aims at identifying and specifying the sources of this bias, in particular in the case of analysis of data issued from outbred populations. Analytical developments were carried out in a backcross situation in order to specify the bias and to propose an algorithm to control it. The outbred population context was studied through simulated data sets in a wide range of situations.The likelihood ratio test was firstly analyzed under the "one QTL" hypothesis in a backcross population. Designs of sib families were then simulated and analyzed using the QTL Map software. On the basis of the theoretical results in backcross, parameters such as the population size, the density of the genetic map, the QTL effect and the true location of the QTL, were taken into account under the "no QTL" and the "one QTL" hypotheses. A combination of two non parametric tests - the Kolmogorov-Smirnov test and the Mann-Whitney-Wilcoxon test - was used in order to identify the parameters that affected the bias and to specify how much they influenced the estimation of QTL location. RESULTS: A theoretical expression of the bias of the estimated QTL location was obtained for a backcross type population. We demonstrated a common source of bias under the "no QTL" and the "one QTL" hypotheses and qualified the possible influence of several parameters. Simulation studies confirmed that the bias exists in outbred populations under both the hypotheses of "no QTL" and "one QTL" on a linkage group. The QTL location was systematically closer to marker locations than expected, particularly in the case of low QTL effect, small population size or low density of markers, i.e. designs with low power. Practical recommendations for experimental designs for QTL detection in outbred populations are given on the basis of this bias quantification. Furthermore, an original algorithm is proposed to adjust the location of a QTL, obtained with interval mapping, which co located with a marker. CONCLUSIONS: Therefore, one should be attentive when one QTL is mapped at the location of one marker, especially under low power conditions.

Validation of carbon isotopologue distribution measurements by GC-MS and application to 13C-metabolic flux analysis of the tricarboxylic acid cycle in Brassica napus leaves.

The estimation of metabolic fluxes in photosynthetic organisms represents an important challenge that has gained interest over the last decade with the development of 13C-Metabolic Flux Analysis at isotopically non-stationary steady-state. This approach requires a high level of accuracy for the measurement of Carbon Isotopologue Distribution in plant metabolites. But this accuracy has still not been evaluated at the isotopologue level for GC-MS, leading to uncertainties for the metabolic fluxes calculated based on these fragments. Here, we developed a workflow to validate the measurements of CIDs from plant metabolites with GC-MS by producing tailor-made E. coli standard extracts harboring a predictable binomial CID for some organic and amino acids. Overall, most of our TMS-derivatives mass fragments were validated with these standards and at natural isotope abundance in plant matrices. Then, we applied this validated MS method to investigate the light/dark regulation of plant TCA cycle by incorporating U-13C-pyruvate to Brassica napus leaf discs. We took advantage of pathway-specific isotopologues/isotopomers observed between two and six hours of labeling to show that the TCA cycle can operate in a cyclic manner under both light and dark conditions. Interestingly, this forward cyclic flux mode has a nearly four-fold higher contribution for pyruvate-to-citrate and pyruvate-to-malate fluxes than the phosphoenolpyruvate carboxylase (PEPc) flux reassimilating carbon derived from some mitochondrial enzymes. The contribution of stored citrate to the mitochondrial TCA cycle activity was also questioned based on dynamics of 13C-enrichment in citrate, glutamate and succinate and variations of citrate total amounts under light and dark conditions. Interestingly, there was a light-dependent 13C-incorporation into glycine and serine showing that decarboxylations from pyruvate dehydrogenase complex and TCA cycle enzymes were actively reassimilated and could represent up to 5% to net photosynthesis.

Suggesting disease associations for overlooked metabolites using literature from metabolic neighbors.

In human health research, metabolic signatures extracted from metabolomics data have a strong added value for stratifying patients and identifying biomarkers. Nevertheless, one of the main challenges is to interpret and relate these lists of discriminant metabolites to pathological mechanisms. This task requires experts to combine their knowledge with information extracted from databases and the scientific literature. However, we show that most compounds (>99%) in the PubChem database lack annotated literature. This dearth of available information can have a direct impact on the interpretation of metabolic signatures, which is often restricted to a subset of significant metabolites. To suggest potential pathological phenotypes related to overlooked metabolites that lack annotated literature, we extend the "guilt-by-association" principle to literature information by using a Bayesian framework. The underlying assumption is that the literature associated with the metabolic neighbors of a compound can provide valuable insights, or an a priori, into its biomedical context. The metabolic neighborhood of a compound can be defined from a metabolic network and correspond to metabolites to which it is connected through biochemical reactions. With the proposed approach, we suggest more than 35,000 associations between 1,047 overlooked metabolites and 3,288 diseases (or disease families). All these newly inferred associations are freely available on the FORUM ftp server (see information at https://github.com/eMetaboHUB/Forum-LiteraturePropagation).

Complex trait subtypes identification using transcriptome profiling reveals an interaction between two QTL affecting adiposity in chicken.

BACKGROUND: Integrative genomics approaches that combine genotyping and transcriptome profiling in segregating populations have been developed to dissect complex traits. The most common approach is to identify genes whose eQTL colocalize with QTL of interest, providing new functional hypothesis about the causative mutation. Another approach includes defining subtypes for a complex trait using transcriptome profiles and then performing QTL mapping using some of these subtypes. This approach can refine some QTL and reveal new ones.In this paper we introduce Factor Analysis for Multiple Testing (FAMT) to define subtypes more accurately and reveal interaction between QTL affecting the same trait. The data used concern hepatic transcriptome profiles for 45 half sib male chicken of a sire known to be heterozygous for a QTL affecting abdominal fatness (AF) on chromosome 5 distal region around 168 cM. RESULTS: Using this methodology which accounts for hidden dependence structure among phenotypes, we identified 688 genes that are significantly correlated to the AF trait and we distinguished 5 subtypes for AF trait, which are not observed with gene lists obtained by classical approaches. After exclusion of one of the two lean bird subtypes, linkage analysis revealed a previously undetected QTL on chromosome 5 around 100 cM. Interestingly, the animals of this subtype presented the same q paternal haplotype at the 168 cM QTL. This result strongly suggests that the two QTL are in interaction. In other words, the "q configuration" at the 168 cM QTL could hide the QTL existence in the proximal region at 100 cM. We further show that the proximal QTL interacts with the previous one detected on the chromosome 5 distal region. CONCLUSION: Our results demonstrate that stratifying genetic population by molecular phenotypes followed by QTL analysis on various subtypes can lead to identification of novel and interacting QTL.

Detection of QTL with effects on osmoregulation capacities in the rainbow trout (Oncorhynchus mykiss).

BACKGROUND: There is increasing evidence that the ability to adapt to seawater in teleost fish is modulated by genetic factors. Most studies have involved the comparison of species or strains and little is known about the genetic architecture of the trait. To address this question, we searched for QTL affecting osmoregulation capacities after transfer to saline water in a nonmigratory captive-bred population of rainbow trout. RESULTS: A QTL design (5 full-sib families, about 200 F2 progeny each) was produced from a cross between F0 grand-parents previously selected during two generations for a high or a low cortisol response after a standardized confinement stress. When fish were about 18 months old (near 204 g body weight), individual progeny were submitted to two successive hyper-osmotic challenges (30 ppt salinity) 14 days apart. Plasma chloride and sodium concentrations were recorded 24 h after each transfer. After the second challenge, fish were sacrificed and a gill index (weight of total gill arches corrected for body weight) was recorded. The genome scan was performed with 196 microsatellites and 85 SNP markers. Unitrait and multiple-trait QTL analyses were carried out on the whole dataset (5 families) through interval mapping methods with the QTLMap software. For post-challenge plasma ion concentrations, significant QTL (P < 0.05) were found on six different linkage groups and highly suggestive ones (P < 0.10) on two additional linkage groups. Most QTL affected concentrations of both chloride and sodium during both challenges, but some were specific to either chloride (2 QTL) or sodium (1 QTL) concentrations. Six QTL (4 significant, 2 suggestive) affecting gill index were discovered. Two were specific to the trait, while the others were also identified as QTL for post-challenge ion concentrations. Altogether, allelic effects were consistent for QTL affecting chloride and sodium concentrations but inconsistent for QTL affecting ion concentrations and gill morphology. There was no systematic lineage effect (grand-parental origin of QTL alleles) on the recorded traits. CONCLUSIONS: For the first time, genomic loci associated with effects on major physiological components of osmotic adaptation to seawater in a nonmigratory fish were revealed. The results pave the way for further deciphering of the complex regulatory mechanisms underlying seawater adaptation and genes involved in osmoregulatory physiology in rainbow trout and other euryhaline fishes.

BioMAJ: a flexible framework for databanks synchronization and processing.

UNLABELLED: Large- and medium-scale computational molecular biology projects require accurate bioinformatics software and numerous heterogeneous biological databanks, which are distributed around the world. BioMAJ provides a flexible, robust, fully automated environment for managing such massive amounts of data. The JAVA application enables automation of the data update cycle process and supervision of the locally mirrored data repository. We have developed workflows that handle some of the most commonly used bioinformatics databases. A set of scripts is also available for post-synchronization data treatment consisting of indexation or format conversion (for NCBI blast, SRS, EMBOSS, GCG, etc.). BioMAJ can be easily extended by personal homemade processing scripts. Source history can be kept via html reports containing statements of locally managed databanks. AVAILABILITY: http://biomaj.genouest.org. BioMAJ is free open software. It is freely available under the CECILL version 2 license.

A fast algorithm for estimating transmission probabilities in QTL detection designs with dense maps.

BACKGROUND: In the case of an autosomal locus, four transmission events from the parents to progeny are possible, specified by the grand parental origin of the alleles inherited by this individual. Computing the probabilities of these transmission events is essential to perform QTL detection methods. RESULTS: A fast algorithm for the estimation of these probabilities conditional to parental phases has been developed. It is adapted to classical QTL detection designs applied to outbred populations, in particular to designs composed of half and/or full sib families. It assumes the absence of interference. CONCLUSION: The theory is fully developed and an example is given.

Graphics processing unit-accelerated quantitative trait Loci detection.

Mapping quantitative trait loci (QTL) using genetic marker information is a time-consuming analysis that has interested the mapping community in recent decades. The increasing amount of genetic marker data allows one to consider ever more precise QTL analyses while increasing the demand for computation. Part of the difficulty of detecting QTLs resides in finding appropriate critical values or threshold values, above which a QTL effect is considered significant. Different approaches exist to determine these thresholds, using either empirical methods or algebraic approximations. In this article, we present a new implementation of existing software, QTLMap, which takes advantage of the data parallel nature of the problem by offsetting heavy computations to a graphics processing unit (GPU). Developments on the GPU were implemented using Cuda technology. This new implementation performs up to 75 times faster than the previous multicore implementation, while maintaining the same results and level of precision (Double Precision) and computing both QTL values and thresholds. This speedup allows one to perform more complex analyses, such as linkage disequilibrium linkage analyses (LDLA) and multiQTL analyses, in a reasonable time frame.

Comparison of the analyses of the XVth QTLMAS common dataset II: QTL analysis.

BACKGROUND: The QTLMAS XVth dataset consisted of the pedigrees, marker genotypes and quantitative trait performances of 2,000 phenotyped animals with a half-sib family structure. The trait was regulated by 8 QTL which display additive, imprinting or epistatic effects. This paper aims at comparing the QTL mapping results obtained by six participants of the workshop. METHODS: Different regression, GBLUP, LASSO and Bayesian methods were applied for QTL detection. The results of these methods are compared based on the number of correctly mapped QTL, the number of false positives, the accuracy of the QTL location and the estimation of the QTL effect. RESULTS: All the simulated QTL, except the interacting QTL on Chr5, were identified by the participants. Depending on the method, 3 to 7 out of the 8 QTL were identified. The distance to the real location and the accuracy of the QTL effect varied to a large extent depending on the methods and complexity of the simulated QTL. CONCLUSIONS: While all methods were fairly efficient in detecting QTL with additive effects, it was clear that for non-additive situations, such as parent-of-origin effects or interactions, the BayesC method gave the best results by detecting 7 out of the 8 simulated QTL, with only two false positives and a good precision (less than 1 cM away on average). Indeed, if LASSO could detect QTL even in complex situations, it was associated with too many false positive results to allow for efficient GWAS. GENMIX, a method based on the phylogenies of local haplotypes, also appeared as a promising approach, which however showed a few more false positives when compared with the BayesC method.

Comparison of analyses of the XVth QTLMAS common dataset III: Genomic Estimations of Breeding Values.

BACKGROUND: The QTLMAS XVth dataset consisted of pedigree, marker genotypes and quantitative trait performances of animals with a sib family structure. Pedigree and genotypes concerned 3,000 progenies among those 2,000 were phenotyped. The trait was regulated by 8 QTLs which displayed additive, imprinting or epistatic effects. The 1,000 unphenotyped progenies were considered as candidates to selection and their Genomic Estimated Breeding Values (GEBV) were evaluated by participants of the XVth QTLMAS workshop. This paper aims at comparing the GEBV estimation results obtained by seven participants to the workshop. METHODS: From the known QTL genotypes of each candidate, two "true" genomic values (TV) were estimated by organizers: the genotypic value of the candidate (TGV) and the expectation of its progeny genotypic values (TBV). GEBV were computed by the participants following different statistical methods: random linear models (including BLUP and Ridge Regression), selection variable techniques (LASSO, Elastic Net) and Bayesian methods. Accuracy was evaluated by the correlation between TV (TGV or TBV) and GEBV presented by participants. Rank correlation of the best 10% of individuals and error in predictions were also evaluated. Bias was tested by regression of TV on GEBV. RESULTS: Large differences between methods were found for all criteria and type of genetic values (TGV, TBV). In general, the criteria ranked consistently methods belonging to the same family. CONCLUSIONS: Bayesian methods - A

XVth QTLMAS: simulated dataset.

BACKGROUND: Our aim was to simulate the data for the QTLMAS2011 workshop following a pig-type family structure under an oligogenic model, each QTL being specific. RESULTS: The population comprised 3000 individuals issued from 20 sires and 200 dams. Within each family, 10 progenies belonged to the experimental population and were assigned phenotypes and marker genotypes and 5 belonged to the selection population, only known on their marker genotypes. A total of 10,000 SNPs carried by 5 chromosomes of 1 Morgan each were simulated. Eight QTL were created (1 quadri-allelic, 2 linked in phase, 2 linked in repulsion, 1 imprinted and 2 epistatic). Random noise was added giving an heritability of 0.30. The marker density, LD and MAF were similar to real life parameters.

QTL detection for a medium density SNP panel: comparison of different LD and LA methods.

BACKGROUND: New molecular technologies allow high throughput genotyping for QTL mapping with dense genetic maps. Therefore, the interest of linkage analysis models against linkage disequilibrium could be questioned. As these two strategies are very sensitive to marker density, experimental design structures, linkage disequilibrium extent and QTL effect, we propose to investigate these parameters effects on QTL detection. METHODS: The XIIIth QTLMAS workshop simulated dataset was analysed using three linkage disequilibrium models and a linkage analysis model. Interval mapping, multivariate and interaction between QTL analyses were performed using QTLMAP. RESULTS: The linkage analysis models identified 13 QTL, from which 10 mapped close of the 18 which were simulated and three other positions being falsely mapped as containing a QTL. Most of the QTLs identified by interval mapping analysis are not clearly detected by any linkage disequilibrium model. In addition, QTL effects are evolving during the time which was not observed using the linkage disequilibrium models. CONCLUSIONS: Our results show that for such a marker density the interval mapping strategy is still better than using the linkage disequilibrium only. While the experimental design structure gives a lot of power to both approaches, the marker density and informativity clearly affect linkage disequilibrium efficiency for QTL detection.