Bees can be colonized by a large diversity of microbes, including beneficial gut symbionts and detrimental pathogens, with implications for bee health. Over the last few years, researchers around the world have collected a huge amount of genomic and transcriptomic data about the composition, genomic content, and gene expression of bee-associated microbial communities. While each of these datasets by itself has provided important insights, the integration of such datasets provides an unprecedented opportunity to obtain a global picture of the microbes associated with bees and their link to bee health. The challenge of such an approach is that datasets are difficult to find within large generalist repositories and are often not readily accessible, which hinders integrative analyses. Here we present a publicly-available online resource, the BeeBiome data portal ( https://www.beebiome.org ), which provides an overview of and easy access to currently available metagenomic datasets involving bee-associated microbes. Currently the data portal contains 33,678 Sequence Read Archive (SRA) experiments for 278 Apoidea hosts. We present the content and functionalities of this portal. By providing access to all bee microbiomes in a single place, with easy filtering on relevant criteria, BeeBiome will allow faster progress of applied and fundamental research on bee biology and health. It should be a useful tool for researchers, academics, funding agencies, and governments, with beneficial impacts for stakeholders.

Abstract The problem of natural language processing over structured data has become a growing research field, both within the relational database and the Semantic Web community, with significant efforts involved in question answering over knowledge graphs (KGQA). However, many of these approaches are either specifically targeted at open-domain question answering using DBpedia, or require large training datasets to translate a natural language question to SPARQL in order to query the knowledge graph. Hence, these approaches often cannot be applied directly to complex scientific datasets where no prior training data is available. In this paper, we focus on the challenges of natural language processing over knowledge graphs of scientific datasets. In particular, we introduce Bio-SODA, a natural language processing engine that does not require training data in the form of question-answer pairs for generating SPARQL queries. Bio-SODA uses a generic graph-based approach for translating user questions to a ranked list of SPARQL candidate queries. Furthermore, Bio-SODA uses a novel ranking algorithm that includes node centrality as a measure of relevance for selecting the best SPARQL candidate query. Our experiments with real-world datasets across several scientific domains, including the official bioinformatics Question Answering over Linked Data (QALD) challenge, as well as the CORDIS dataset of European projects, show that Bio-SODA outperforms publicly available KGQA systems by an F1-score of least 20% and by an even higher factor on more complex bioinformatics datasets. Finally, we introduce Bio-SODA UX, a graphical user interface designed to assist users in the exploration of large knowledge graphs and in dynamically disambiguating natural language questions that target the data available in these graphs.

Many genes have nycthemeral rhythms of expression, i.e. a 24-hours periodic variation, at either mRNA or protein level or both, and most rhythmic genes are tissue-specific. Here, we investigate and discuss the evolutionary origins of rhythms in gene expression. Our results suggest that rhythmicity of protein expression could have been favored by selection to minimize costs. Trends are consistent in bacteria, plants and animals, and are also supported by tissue-specific patterns in mouse. Unlike for protein level, cost cannot explain rhythm at the RNA level. We suggest that instead it allows to periodically reduce expression noise. Noise control had the strongest support in mouse, with limited evidence in other species. We have also found that genes under stronger purifying selection are rhythmically expressed at the mRNA level, and we propose that this is because they are noise sensitive genes. Finally, the adaptive role of rhythmic expression is supported by rhythmic genes being highly expressed yet tissue-specific. This provides a good evolutionary explanation for the observation that nycthemeral rhythms are often tissue-specific.

The problem of natural language processing over structured data has become a growing research field, both within the relational database and the Semantic Web community, with significant efforts involved in question answering over knowledge graphs (KGQA). However, many of these approaches are either specifically targeted at open-domain question answering using DBpedia, or require large training datasets to translate a natural language question to SPARQL in order to query the knowledge graph. Hence, these approaches often cannot be applied directly to complex scientific datasets where no prior training data is available.

MOTIVATION: Assigning new sequences to known protein families and subfamilies is a prerequisite for many functional, comparative and evolutionary genomics analyses. Such assignment is commonly achieved by looking for the closest sequence in a reference database, using a method such as BLAST. However, ignoring the gene phylogeny can be misleading because a query sequence does not necessarily belong to the same subfamily as its closest sequence. For example, a hemoglobin which branched out prior to the hemoglobin alpha/beta duplication could be closest to a hemoglobin alpha or beta sequence, whereas it is neither. To overcome this problem, phylogeny-driven tools have emerged but rely on gene trees, whose inference is computationally expensive. RESULTS: Here, we first show that in multiple animal and plant datasets, 18-62% of assignments by closest sequence are misassigned, typically to an over-specific subfamily. Then, we introduce OMAmer, a novel alignment-free protein subfamily assignment method, which limits over-specific subfamily assignments and is suited to phylogenomic databases with thousands of genomes. OMAmer is based on an innovative method using evolutionarily informed k-mers for alignment-free mapping to ancestral protein subfamilies. Whilst able to reject non-homologous family-level assignments, we show that OMAmer provides better and quicker subfamily-level assignments than approaches relying on the closest sequence, whether inferred exactly by Smith-Waterman or by the fast heuristic DIAMOND. AVAILABILITYAND IMPLEMENTATION: OMAmer is available from the Python Package Index (as omamer), with the source code and a precomputed database available at https://github.com/DessimozLab/omamer. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The nycthemeral transcriptome embodies all genes displaying a rhythmic variation of their mRNAs periodically every 24 hours, including but not restricted to circadian genes. In this study, we show that the nycthemeral rhythmicity at the gene expression level is biologically functional and that this functionality is more conserved between orthologous genes than between random genes. We used this conservation of the rhythmic expression to assess the ability of seven methods (ARSER, Lomb Scargle, RAIN, JTK, empirical-JTK, GeneCycle, and meta2d) to detect rhythmic signal in gene expression. We have contrasted them to a naive method, not based on rhythmic parameters. By taking into account the tissue-specificity of rhythmic gene expression and different species comparisons, we show that no method is strongly favored. The results show that these methods designed for rhythm detection, in addition to having quite similar performances, are consistent only among genes with a strong rhythm signal. Rhythmic genes defined with a standard p-value threshold of 0.01 for instance, could include genes whose rhythmicity is biologically irrelevant. Although these results were dependent on the datasets used and the evolutionary distance between the species compared, we call for caution about the results of studies reporting or using large sets of rhythmic genes. Furthermore, given the analysis of the behaviors of the methods on real and randomized data, we recommend using primarily ARS, empJTK, or GeneCycle, which verify expectations of a classical distribution of p-values. Experimental design should also take into account the circumstances under which the methods seem more efficient, such as giving priority to biological replicates over the number of time-points, or to the number of time-points over the quality of the technique (microarray vs RNAseq). GeneCycle, and to a lesser extent empirical-JTK, might be the most robust method when applied to weakly informative datasets. Finally, our analyzes suggest that rhythmic genes are mainly highly expressed genes.

The evolutionarily conserved nature of the few well-known anti-aging interventions that affect lifespan, such as caloric restriction, suggests that aging-related research in model organisms is directly relevant to human aging.Since human lifespan is a complex trait, a systems-level approach will contribute to a more comprehensive understanding of the underlying aging landscape.Here, we integrate evolutionary and functional information of normal aging across human and model organisms at three levels: gene-level, process-level, and network-level.We identify evolutionarily conserved modules of normal aging across diverse taxa, and notably show proteostasis to be conserved in normal aging.Additionally, we find that mechanisms related to protein quality control network are enriched for genes harboring genetic variants associated with 22 age-related human traits and associated to caloric restriction.These results demonstrate that a systems-level approach, combined with evolutionary conservation, allows the detection of candidate aging genes and pathways relevant to human normal aging. Author summaryAging leads to changes in the activity of genes, but these changes are much more subtle than those between health and disease.This makes finding the genes involved in aging difficult.We have combined two ways to improve information on the involvement of genes in aging.First, important gene activity tends to be conserved in evolution, so we looked for genes whose changes in aging were conserved between distant animal species.Second, genes act in groups, so we looked for changes which affect groups of genes which act together.We were thus able to detect with good confidence subtle changes, which we could confirm with results of experiments modifying diet, and from large genetic scans in humans.

One of the major properties of genes is their expression pattern. Notably, genes are often classified as tissue specific or housekeeping. This property is of interest to molecular evolution as an explanatory factor of, e.g. evolutionary rate, as well as a functional feature which may in itself evolve. While many different methods of measuring tissue specificity have been proposed and used for such studies, there has been no comparison or benchmarking of these methods to our knowledge, and little justification of their use. In this study, we compare nine measures of tissue specificity. Most methods were established for ESTs and microarrays, and several were later adapted to RNA-seq. We analyse their capacity to distinguish gene categories, their robustness to the choice and number of tissues used and their capture of evolutionary conservation signal.

Motivation: Codon models are widely used to identify the signature of selection at the molecular level and to test for changes in selective pressure during the evolution of genes encoding proteins. The large size of the state space of the Markov processes used to model codon evolution makes it difficult to use these models with large biological datasets. We propose here to use state aggregation to reduce the state space of codon models and, thus, improve the computational performance of likelihood estimation on these models. Results: We show that this heuristic speeds up the computations of the M0 and branch-site models up to 6.8 times. We also show through simulations that state aggregation does not introduce a detectable bias. We analyzed a real dataset and show that aggregation provides highly correlated predictions compared to the full likelihood computations. Finally, state aggregation is a very general approach and can be applied to any continuous-time Markov process-based model with large state space, such as amino acid and coevolution models. We therefore discuss different ways to apply state aggregation to Markov models used in phylogenetics. Availability and Implementation: The heuristic is implemented in the godon package (https://bitbucket.org/Davydov/godon) and in a version of FastCodeML (https://gitlab.isb-sib.ch/phylo/fastcodeml). Contact: [email protected] Supplementary Information: Supplementary data are available at Bioinformatics online.

Bioinformatics (2014)30(10),1392–1399 doi: 10.1093/bioinformatics/btu027 The authors of the above article wish to alert the reader to the following FTP link, as the link given in the published paper is no longer active: Availability and Implementation: The R implementation of IQRray is available at:ftp://ftp.bgee.org/general/IQRray.R

The ortholog conjecture implies that functional similarity between orthologous genes is higher than between paralogs. It has been supported using levels of expression and Gene Ontology term analysis, although the evidence was rather weak and there were also conflicting reports. In this study on 12 species we provide strong evidence of high conservation in tissue-specificity between orthologs, in contrast to low conservation between within-species paralogs. This allows us to shed a new light on the evolution of gene expression patterns. While there have been several studies of the correlation of expression between species, little is known about the evolution of tissue-specificity itself. Ortholog tissue-specificity is strongly conserved between all tetrapod species, with the lowest Pearson correlation between mouse and frog at r = 0.66. Tissue-specificity correlation decreases strongly with divergence time. Paralogs in human show much lower conservation, even for recent Primate-specific paralogs. When both paralogs from ancient whole genome duplication tissue-specific paralogs are tissue-specific, it is often to different tissues, while other tissue-specific paralogs are mostly specific to the same tissue. The same patterns are observed using human or mouse as focal species, and are robust to choices of datasets and of thresholds. Our results support the following model of evolution: in the absence of duplication, tissue-specificity evolves slowly, and tissue-specific genes do not change their main tissue of expression; after small-scale duplication the less expressed paralog loses the ancestral specificity, leading to an immediate difference between paralogs; over time, both paralogs become more broadly expressed, but remain poorly correlated. Finally, there is a small number of paralog pairs which stay tissue-specific with the same main tissue of expression, for at least 300 million years.

MOTIVATION: Microarray results accumulated in public repositories are widely reused in meta-analytical studies and secondary databases. The quality of the data obtained with this technology varies from experiment to experiment, and an efficient method for quality assessment is necessary to ensure their reliability. RESULTS: The lack of a good benchmark has hampered evaluation of existing methods for quality control. In this study, we propose a new independent quality metric that is based on evolutionary conservation of expression profiles. We show, using 11 large organ-specific datasets, that IQRray, a new quality metrics developed by us, exhibits the highest correlation with this reference metric, among 14 metrics tested. IQRray outperforms other methods in identification of poor quality arrays in datasets composed of arrays from many independent experiments. In contrast, the performance of methods designed for detecting outliers in a single experiment like Normalized Unscaled Standard Error and Relative Log Expression was low because of the inability of these methods to detect datasets containing only low-quality arrays and because the scores cannot be directly compared between experiments. AVAILABILITY AND IMPLEMENTATION: The R implementation of IQRray is available at: ftp://lausanne.isb-sib.ch/pub/databases/Bgee/general/IQRray.R. CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

UNLABELLED: Given the rapid increase of species with a sequenced genome, the need to identify orthologous genes between them has emerged as a central bioinformatics task. Many different methods exist for orthology detection, which makes it difficult to decide which one to choose for a particular application. Here, we review the latest developments and issues in the orthology field, and summarize the most recent results reported at the third 'Quest for Orthologs' meeting. We focus on community efforts such as the adoption of reference proteomes, standard file formats and benchmarking. Progress in these areas is good, and they are already beneficial to both orthology consumers and providers. However, a major current issue is that the massive increase in complete proteomes poses computational challenges to many of the ortholog database providers, as most orthology inference algorithms scale at least quadratically with the number of proteomes. The Quest for Orthologs consortium is an open community with a number of working groups that join efforts to enhance various aspects of orthology analysis, such as defining standard formats and datasets, documenting community resources and benchmarking. AVAILABILITY AND IMPLEMENTATION: All such materials are available at http://questfororthologs.org.

MOTIVATION: The detection of positive selection is widely used to study gene and genome evolution, but its application remains limited by the high computational cost of existing implementations. We present a series of computational optimizations for more efficient estimation of the likelihood function on large-scale phylogenetic problems. We illustrate our approach using the branch-site model of codon evolution. RESULTS: We introduce novel optimization techniques that substantially outperform both CodeML from the PAML package and our previously optimized sequential version SlimCodeML. These techniques can also be applied to other likelihood-based phylogeny software. Our implementation scales well for large numbers of codons and/or species. It can therefore analyse substantially larger datasets than CodeML. We evaluated FastCodeML on different platforms and measured average sequential speedups of FastCodeML (single-threaded) versus CodeML of up to 5.8, average speedups of FastCodeML (multi-threaded) versus CodeML on a single node (shared memory) of up to 36.9 for 12 CPU cores, and average speedups of the distributed FastCodeML versus CodeML of up to 170.9 on eight nodes (96 CPU cores in total). AVAILABILITY AND IMPLEMENTATION: ftp://ftp.vital-it.ch/tools/FastCodeML/ CONTACT: [email protected] or [email protected].

The influence of different parameters, from gene size to expression levels, on the evolution of proteins has been previously studied mostly in yeast [ 1 ] and Drosophila [ 2 ]. The main feature which has been found to explain protein evolutionary rate was the level of gene expression, especially in yeast. Here we investigate these relations further, and extend them to mammals, especially taking in account gene expression in different organs. For expression we used the RNA-seq data from ENCODE [ 3 ] for 22 different tissues of mouse. We used ENCODE data to define which transcript is used as reference to compute features such as gene length or intron number. The relation between evolutionary rate and six gene features: gene expression, gene expression specificity, intron number, intron length, protein length and GC% content were analyzed. We use partial correlation to take into account dependencies between them. We find strong differences between tissues in the impact of expression on evolutionary rate (Figure 1 and http://f1000.com/posters/browse/summary/1094165 ). Over all tissues, an interesting result is that evolutionary rate shows no strong correlation with expression level in mouse if corrected for other parameters. Partial correlations of gene parameters (mouse data) with Pearson correlation coefficient. The width of connections shows the strength of the correlations. Only correlations with p-value > 0.001 are shown. Positive correlations are represented in yellow and negative in blue. Tissues were ordered according to correlation with evolutionary rate. Dependencies between gene features need to be taken into account for an unbiased view of gene evolution. Overall results are consistent with those in Drosophila [ 2 ]. We find important differences between tissues in the relation between expression and evolutionary rate, especially for the central nervous system and testis.

MOTIVATION: Most anatomical ontologies are species-specific, whereas a framework for comparative studies is needed. We describe the vertebrate Homologous Organs Groups ontology, vHOG, used to compare expression patterns between species. RESULTS: vHOG is a multispecies anatomical ontology for the vertebrate lineage. It is based on the HOGs used in the Bgee database of gene expression evolution. vHOG version 1.4 includes 1184 terms, follows OBO principles and is based on the Common Anatomy Reference Ontology (CARO). vHOG only describes structures with historical homology relations between model vertebrate species. The mapping to species-specific anatomical ontologies is provided as a separate file, so that no homology hypothesis is stated within the ontology itself. Each mapping has been manually reviewed, and we provide support codes and references when available. AVAILABILITY AND IMPLEMENTATION: vHOG is available from the Bgee download site (http://bgee.unil.ch/), as well as from the OBO Foundry and the NCBO Bioportal websites. CONTACT: [email protected]; [email protected].

MOTIVATION: Comparative analyses of gene expression data from different species have become an important component of the study of molecular evolution. Thus methods are needed to estimate evolutionary distances between expression profiles, as well as a neutral reference to estimate selective pressure. Divergence between expression profiles of homologous genes is often calculated with Pearson's or Euclidean distance. Neutral divergence is usually inferred from randomized data. Despite being widely used, neither of these two steps has been well studied. Here, we analyze these methods formally and on real data, highlight their limitations and propose improvements. RESULTS: It has been demonstrated that Pearson's distance, in contrast to Euclidean distance, leads to underestimation of the expression similarity between homologous genes with a conserved uniform pattern of expression. Here, we first extend this study to genes with conserved, but specific pattern of expression. Surprisingly, we find that both Pearson's and Euclidean distances used as a measure of expression similarity between genes depend on the expression specificity of those genes. We also show that the Euclidean distance depends strongly on data normalization. Next, we show that the randomization procedure that is widely used to estimate the rate of neutral evolution is biased when broadly expressed genes are abundant in the data. To overcome this problem, we propose a novel randomization procedure that is unbiased with respect to expression profiles present in the datasets. Applying our method to the mouse and human gene expression data suggests significant gene expression conservation between these species.

The function of most proteins is not determined experimentally, but is extrapolated from homologs. According to the "ortholog conjecture", or standard model of phylogenomics, protein function changes rapidly after duplication, leading to paralogs with different functions, while orthologs retain the ancestral function. We report here that a comparison of experimentally supported functional annotations among homologs from 13 genomes mostly supports this model. We show that to analyze GO annotation effectively, several confounding factors need to be controlled: authorship bias, variation of GO term frequency among species, variation of background similarity among species pairs, and propagated annotation bias. After controlling for these biases, we observe that orthologs have generally more similar functional annotations than paralogs. This is especially strong for sub-cellular localization. We observe only a weak decrease in functional similarity with increasing sequence divergence. These findings hold over a large diversity of species; notably orthologs from model organisms such as E. coli, yeast or mouse have conserved function with human proteins.

Phylogenomic databases provide orthology predictions for species with fully sequenced genomes. Although the goal seems well-defined, the content of these databases differs greatly. Seven ortholog databases (Ensembl Compara, eggNOG, HOGENOM, InParanoid, OMA, OrthoDB, Panther) were compared on the basis of reference trees. For three well-conserved protein families, we observed a generally high specificity of orthology assignments for these databases. We show that differences in the completeness of predicted gene relationships and in the phylogenetic information are, for the great majority, not due to the methods used, but to differences in the underlying database concepts. According to our metrics, none of the databases provides a fully correct and comprehensive protein classification. Our results provide a framework for meaningful and systematic comparisons of phylogenomic databases. In the future, a sustainable set of 'Gold standard' phylogenetic trees could provide a robust method for phylogenomic databases to assess their current quality status, measure changes following new database releases and diagnose improvements subsequent to an upgrade of the analysis procedure.

Despite the common assumption that orthologs usually share the same function, there have been various reports of divergence between orthologs, even among species as close as mammals. The comparison of mouse and human is of special interest, because mouse is often used as a model organism to understand human biology. We review the literature on evidence for divergence between human and mouse orthologous genes, and discuss it in the context of biomedical research.

MOTIVATION: The anatomy of model species is described in ontologies, which are used to standardize the annotations of experimental data, such as gene expression patterns. To compare such data between species, we need to establish relations between ontologies describing different species. RESULTS: We present a new algorithm, and its implementation in the software Homolonto, to create new relationships between anatomical ontologies, based on the homology concept. Homolonto uses a supervised ontology alignment approach. Several alignments can be merged, forming homology groups. We also present an algorithm to generate relationships between these homology groups. This has been used to build a multi-species ontology, for the database of gene expression evolution Bgee. AVAILABILITY: download section of the Bgee website http://bgee.unil.ch/

Abstract *To whom correspondence should be addressed. 3Present address: Laboratoire de Biologie Moléculaire et Cellulaire, Ecole Normale Supérieure de Lyon, 6 allée d’Italie, 69364 Lyon cedex 07, France, [email protected] Summary: RRTree is a user-friendly program for comparing substitution rates between lineages of protein or DNA sequences, relative to an outgroup, through relative rate tests. Genetic diversity is taken into account through use of several sequences, and phylogenetic relations are integrated by topological weighting. Availability: The ANSI C source code of RRTree, and compiled versions for Macintosh, MS-DOS/Windows, SUN Solaris, and CGI, are freely available at http://pbil.univ-lyon1.fr/software/rrtree.html Contact: [email protected]

SUMMARY: JaDis is a Java application for computing evolutionary distances between nucleic acid sequences and G+C base frequencies. It allows specific comparison of coding sequences, of non-coding sequences or of a non-coding sequence with coding sequences. AVAILABILITY: http://pbil.univ-lyon1.fr/software/jadis.html