Converting disease maps into heavyweight ontologies: general methodology and application to Alzheimer's disease.

Omics technologies offer great promises for improving our understanding of diseases. The integration and interpretation of such data pose major challenges, calling for adequate knowledge models. Disease maps provide curated knowledge about disorders' pathophysiology at the molecular level adapted to omics measurements. However, the expressiveness of disease maps could be increased to help in avoiding ambiguities and misinterpretations and to reinforce their interoperability with other knowledge resources. Ontology is an adequate framework to overcome this limitation, through their axiomatic definitions and logical reasoning properties. We introduce the Disease Map Ontology (DMO), an ontological upper model based on systems biology terms. We then propose to apply DMO to Alzheimer's disease (AD). Specifically, we use it to drive the conversion of AlzPathway, a disease map devoted to AD, into a formal ontology: Alzheimer DMO. We demonstrate that it allows one to deal with issues related to redundancy, naming, consistency, process classification and pathway relationships. Furthermore, we show that it can store and manage multi-omics data. Finally, we expand the model using elements from other resources, such as clinical features contained in the AD Ontology, resulting in an enriched model called ADMO-plus. The current versions of DMO, ADMO and ADMO-plus are freely available at http://bioportal.bioontology.org/ontologies/ADMO.

Plasma microRNA signature in presymptomatic and symptomatic subjects with C9orf72-associated frontotemporal dementia and amyotrophic lateral sclerosis.

OBJECTIVE: To identify potential biomarkers of preclinical and clinical progression in chromosome 9 open reading frame 72 gene (C9orf72)-associated disease by assessing the expression levels of plasma microRNAs (miRNAs) in C9orf72 patients and presymptomatic carriers. METHODS: The PREV-DEMALS study is a prospective study including 22 C9orf72 patients, 45 presymptomatic C9orf72 mutation carriers and 43 controls. We assessed the expression levels of 2576 miRNAs, among which 589 were above noise level, in plasma samples of all participants using RNA sequencing. The expression levels of the differentially expressed miRNAs between patients, presymptomatic carriers and controls were further used to build logistic regression classifiers. RESULTS: Four miRNAs were differentially expressed between patients and controls: miR-34a-5p and miR-345-5p were overexpressed, while miR-200c-3p and miR-10a-3p were underexpressed in patients. MiR-34a-5p was also overexpressed in presymptomatic carriers compared with healthy controls, suggesting that miR-34a-5p expression is deregulated in cases with C9orf72 mutation. Moreover, miR-345-5p was also overexpressed in patients compared with presymptomatic carriers, which supports the correlation of miR-345-5p expression with the progression of C9orf72-associated disease. Together, miR-200c-3p and miR-10a-3p underexpression might be associated with full-blown disease. Four presymptomatic subjects in transitional/prodromal stage, close to the disease conversion, exhibited a stronger similarity with the expression levels of patients. CONCLUSIONS: We identified a signature of four miRNAs differentially expressed in plasma between clinical conditions that have potential to represent progression biomarkers for C9orf72-associated frontotemporal dementia and amyotrophic lateral sclerosis. This study suggests that dysregulation of miRNAs is dynamically altered throughout neurodegenerative diseases progression, and can be detectable even long before clinical onset. TRIAL REGISTRATION NUMBER: NCT02590276.

Diet-Induced Dysbiosis and Genetic Background Synergize With Cystic Fibrosis Transmembrane Conductance Regulator Deficiency to Promote Cholangiopathy in Mice.

The most typical expression of cystic fibrosis (CF)-related liver disease is a cholangiopathy that can progress to cirrhosis. We aimed to determine the potential impact of environmental and genetic factors on the development of CF-related cholangiopathy in mice. Cystic fibrosis transmembrane conductance regulator (Cftr)-/- mice and Cftr +/+ littermates in a congenic C57BL/6J background were fed a high medium-chain triglyceride (MCT) diet. Liver histopathology, fecal microbiota, intestinal inflammation and barrier function, bile acid homeostasis, and liver transcriptome were analyzed in 3-month-old males. Subsequently, MCT diet was changed for chow with polyethylene glycol (PEG) and the genetic background for a mixed C57BL/6J;129/Ola background (resulting from three backcrosses), to test their effect on phenotype. C57BL/6J Cftr -/- mice on an MCT diet developed cholangiopathy features that were associated with dysbiosis, primarily Escherichia coli enrichment, and low-grade intestinal inflammation. Compared with Cftr +/+ littermates, they displayed increased intestinal permeability and a lack of secondary bile acids together with a low expression of ileal bile acid transporters. Dietary-induced (chow with PEG) changes in gut microbiota composition largely prevented the development of cholangiopathy in Cftr -/- mice. Regardless of Cftr status, mice in a mixed C57BL/6J;129/Ola background developed fatty liver under an MCT diet. The Cftr -/- mice in the mixed background showed no cholangiopathy, which was not explained by a difference in gut microbiota or intestinal permeability, compared with congenic mice. Transcriptomic analysis of the liver revealed differential expression, notably of immune-related genes, in mice of the congenic versus mixed background. In conclusion, our findings suggest that CFTR deficiency causes abnormal intestinal permeability, which, combined with diet-induced dysbiosis and immune-related genetic susceptibility, promotes CF-related cholangiopathy.

Oligodendrocyte precursor survival and differentiation requires chromatin remodeling by Chd7 and Chd8.

Oligodendrocyte precursor cells (OPCs) constitute the main proliferative cells in the adult brain, and deregulation of OPC proliferation-differentiation balance results in either glioma formation or defective adaptive (re)myelination. OPC differentiation requires significant genetic reprogramming, implicating chromatin remodeling. Mounting evidence indicates that chromatin remodelers play important roles during normal development and their mutations are associated with neurodevelopmental defects, with CHD7 haploinsuficiency being the cause of CHARGE syndrome and CHD8 being one of the strongest autism spectrum disorder (ASD) high-risk-associated genes. Herein, we report on uncharacterized functions of the chromatin remodelers Chd7 and Chd8 in OPCs. Their OPC-chromatin binding profile, combined with transcriptome and chromatin accessibility analyses of Chd7-deleted OPCs, demonstrates that Chd7 protects nonproliferative OPCs from apoptosis by chromatin closing and transcriptional repression of p53 Furthermore, Chd7 controls OPC differentiation through chromatin opening and transcriptional activation of key regulators, including Sox10, Nkx2.2, and Gpr17 However, Chd7 is dispensable for oligodendrocyte stage progression, consistent with Chd8 compensatory function, as suggested by their common chromatin-binding profiles and genetic interaction. Finally, CHD7 and CHD8 bind in OPCs to a majority of ASD risk-associated genes, suggesting an implication of oligodendrocyte lineage cells in ASD neurological defects. Our results thus offer new avenues to understand and modulate the CHD7 and CHD8 functions in normal development and disease.

A strategy for multimodal data integration: application to biomarkers identification in spinocerebellar ataxia.

The growing number of modalities (e.g. multi-omics, imaging and clinical data) characterizing a given disease provides physicians and statisticians with complementary facets reflecting the disease process but emphasizes the need for novel statistical methods of data analysis able to unify these views. Such data sets are indeed intrinsically structured in blocks, where each block represents a set of variables observed on a group of individuals. Therefore, classical statistical tools cannot be applied without altering their organization, with the risk of information loss. Regularized generalized canonical correlation analysis (RGCCA) and its sparse generalized canonical correlation analysis (SGCCA) counterpart are component-based methods for exploratory analyses of data sets structured in blocks of variables. Rather than operating sequentially on parts of the measurements, the RGCCA/SGCCA-based integrative analysis method aims at summarizing the relevant information between and within the blocks. It processes a priori information defining which blocks are supposed to be linked to one another, thus reflecting hypotheses about the biology underlying the data blocks. It also requires the setting of extra parameters that need to be carefully adjusted.Here, we provide practical guidelines for the use of RGCCA/SGCCA. We also illustrate the flexibility and usefulness of RGCCA/SGCCA on a unique cohort of patients with four genetic subtypes of spinocerebellar ataxia, in which we obtained multiple data sets from brain volumetry and magnetic resonance spectroscopy, and metabolomic and lipidomic analyses. As a first step toward the extraction of multimodal biomarkers, and through the reduction to a few meaningful components and the visualization of relevant variables, we identified possible markers of disease progression.

Clinical-genetic model predicts incident impulse control disorders in Parkinson's disease.

OBJECTIVES: Impulse control disorders (ICD) are commonly associated with dopamine replacement therapy (DRT) in patients with Parkinson's disease (PD). Our aims were to estimate ICD heritability and to predict ICD by a candidate genetic multivariable panel in patients with PD. METHODS: Data from de novo patients with PD, drug-naïve and free of ICD behaviour at baseline, were obtained from the Parkinson's Progression Markers Initiative cohort. Incident ICD behaviour was defined as positive score on the Questionnaire for Impulsive-Compulsive Disorders in PD. ICD heritability was estimated by restricted maximum likelihood analysis on whole exome sequencing data. 13 candidate variants were selected from the DRD2, DRD3, DAT1, COMT, DDC, GRIN2B, ADRA2C, SERT, TPH2, HTR2A, OPRK1 and OPRM1 genes. ICD prediction was evaluated by the area under the curve (AUC) of receiver operating characteristic (ROC) curves. RESULTS: Among 276 patients with PD included in the analysis, 86% started DRT, 40% were on dopamine agonists (DA), 19% reported incident ICD behaviour during follow-up. We found heritability of this symptom to be 57%. Adding genotypes from the 13 candidate variants significantly increased ICD predictability (AUC=76%, 95% CI (70% to 83%)) compared to prediction based on clinical variables only (AUC=65%, 95% CI (58% to 73%), p=0.002). The clinical-genetic prediction model reached highest accuracy in patients initiating DA therapy (AUC=87%, 95% CI (80% to 93%)). OPRK1, HTR2A and DDC genotypes were the strongest genetic predictive factors. CONCLUSIONS: Our results show that adding a candidate genetic panel increases ICD predictability, suggesting potential for developing clinical-genetic models to identify patients with PD at increased risk of ICD development and guide DRT management.

Genome-wide replication landscape of Candida glabrata.

BACKGROUND: The opportunistic pathogen Candida glabrata is a member of the Saccharomycetaceae yeasts. Like its close relative Saccharomyces cerevisiae, it underwent a whole-genome duplication followed by an extensive loss of genes. Its genome contains a large number of very long tandem repeats, called megasatellites. In order to determine the whole replication program of the C. glabrata genome and its general chromosomal organization, we used deep-sequencing and chromosome conformation capture experiments. RESULTS: We identified 253 replication fork origins, genome wide. Centromeres, HML and HMR loci, and most histone genes are replicated early, whereas natural chromosomal breakpoints are located in late-replicating regions. In addition, 275 autonomously replicating sequences (ARS) were identified during ARS-capture experiments, and their relative fitness was determined during growth competition. Analysis of ARSs allowed us to identify a 17-bp consensus, similar to the S. cerevisiae ARS consensus sequence but slightly more constrained. Megasatellites are not in close proximity to replication origins or termini. Using chromosome conformation capture, we also show that early origins tend to cluster whereas non-subtelomeric megasatellites do not cluster in the yeast nucleus. CONCLUSIONS: Despite a shorter cell cycle, the C. glabrata replication program shares unexpected striking similarities to S. cerevisiae, in spite of their large evolutionary distance and the presence of highly repetitive large tandem repeats in C. glabrata. No correlation could be found between the replication program and megasatellites, suggesting that their formation and propagation might not be directly caused by replication fork initiation or termination.

Streptococcus agalactiae clones infecting humans were selected and fixed through the extensive use of tetracycline.

Streptococcus agalactiae (Group B Streptococcus, GBS) is a commensal of the digestive and genitourinary tracts of humans that emerged as the leading cause of bacterial neonatal infections in Europe and North America during the 1960s. Due to the lack of epidemiological and genomic data, the reasons for this emergence are unknown. Here we show by comparative genome analysis and phylogenetic reconstruction of 229 isolates that the rise of human GBS infections corresponds to the selection and worldwide dissemination of only a few clones. The parallel expansion of the clones is preceded by the insertion of integrative and conjugative elements conferring tetracycline resistance (TcR). Thus, we propose that the use of tetracycline from 1948 onwards led in humans to the complete replacement of a diverse GBS population by only few TcR clones particularly well adapted to their host, causing the observed emergence of GBS diseases in neonates.

Comparison of widely used Listeria monocytogenes strains EGD, 10403S, and EGD-e highlights genomic variations underlying differences in pathogenicity.

For nearly 3 decades, listeriologists and immunologists have used mainly three strains of the same serovar (1/2a) to analyze the virulence of the bacterial pathogen Listeria monocytogenes. The genomes of two of these strains, EGD-e and 10403S, were released in 2001 and 2008, respectively. Here we report the genome sequence of the third reference strain, EGD, and extensive genomic and phenotypic comparisons of the three strains. Strikingly, EGD-e is genetically highly distinct from EGD (29,016 single nucleotide polymorphisms [SNPs]) and 10403S (30,296 SNPs), and is more related to serovar 1/2c than 1/2a strains. We also found that while EGD and 10403S strains are genetically very close (317 SNPs), EGD has a point mutation in the transcriptional regulator PrfA (PrfA*), leading to constitutive expression of several major virulence genes. We generated an EGD-e PrfA* mutant and showed that EGD behaves like this strain in vitro, with slower growth in broth and higher invasiveness in human cells than those of EGD-e and 10403S. In contrast, bacterial counts in blood, liver, and spleen during infection in mice revealed that EGD and 10403S are less virulent than EGD-e, which is itself less virulent than EGD-e PrfA*. Thus, constitutive expression of PrfA-regulated virulence genes does not appear to provide a significant advantage to the EGD strain during infection in vivo, highlighting the fact that in vitro invasion assays are not sufficient for evaluating the pathogenic potential of L. monocytogenes strains. Together, our results pave the way for deciphering unexplained differences or discrepancies in experiments using different L. monocytogenes strains. IMPORTANCE Over the past 3 decades, Listeria has become a model organism for host-pathogen interactions, leading to critical discoveries in a broad range of fields, including bacterial gene regulation, cell biology, and bacterial pathophysiology. Scientists studying Listeria use primarily three pathogenic strains: EGD, EGD-e, and 10403S. Despite many studies on EGD, it is the only one of the three strains whose genome has not been sequenced. Here we report the sequence of its genome and a series of important genomic and phenotypic differences between the three strains, in particular, a critical mutation in EGD's PrfA, the main regulator of Listeria virulence. Our results show that the three strains display differences which may play an important role in the virulence differences observed between the strains. Our findings will be of critical relevance to listeriologists and immunologists who have used or may use Listeria as a tool to study the pathophysiology of listeriosis and immune responses.

SynTView - an interactive multi-view genome browser for next-generation comparative microorganism genomics.

BACKGROUND: Dynamic visualisation interfaces are required to explore the multiple microbial genome data now available, especially those obtained by high-throughput sequencing - a.k.a. "Next-Generation Sequencing" (NGS) - technologies; they would also be useful for "standard" annotated genomes whose chromosome organizations may be compared. Although various software systems are available, few offer an optimal combination of feature-rich capabilities, non-static user interfaces and multi-genome data handling. RESULTS: We developed SynTView, a comparative and interactive viewer for microbial genomes, designed to run as either a web-based tool (Flash technology) or a desktop application (AIR environment). The basis of the program is a generic genome browser with sub-maps holding information about genomic objects (annotations). The software is characterised by the presentation of syntenic organisations of microbial genomes and the visualisation of polymorphism data (typically Single Nucleotide Polymorphisms - SNPs) along these genomes; these features are accessible to the user in an integrated way. A variety of specialised views are available and are all dynamically inter-connected (including linear and circular multi-genome representations, dot plots, phylogenetic profiles, SNP density maps, and more). SynTView is not linked to any particular database, allowing the user to plug his own data into the system seamlessly, and use external web services for added functionalities. SynTView has now been used in several genome sequencing projects to help biologists make sense out of huge data sets. CONCLUSIONS: The most important assets of SynTView are: (i) the interactivity due to the Flash technology; (ii) the capabilities for dynamic interaction between many specialised views; and (iii) the flexibility allowing various user data sets to be integrated. It can thus be used to investigate massive amounts of information efficiently at the chromosome level. This innovative approach to data exploration could not be achieved with most existing genome browsers, which are more static and/or do not offer multiple views of multiple genomes. Documentation, tutorials and demonstration sites are available at the URL: http://genopole.pasteur.fr/SynTView.

Draft Genome Sequence of a Clinical Strain of Yersinia enterocolitica (IP10393) of Bioserotype 4/O:3 from France.

We sequenced the genome of a clinical isolate of Yersinia enterocolitica (IP10393) from France. This strain belongs to bioserotype 4/O:3, which is the most common pathogenic subgroup worldwide. The draft genome has a size of 4,463,212 bp and a G+C content of 47.0%, and it is predicted to contain 4,181 coding sequences.

Evaluation of high-throughput sequencing for identifying known and unknown viruses in biological samples.

High-throughput sequencing furnishes a large number of short sequence reads from uncloned DNA and has rapidly become a major tool for identifying viruses in biological samples, and in particular when the target sequence is undefined. In this study, we assessed the analytical sensitivity of a pipeline for detection of viruses in biological samples based on either the Roche-454 genome sequencer or Illumina genome analyzer platforms. We sequenced biological samples artificially spiked with a wide range of viruses with genomes composed of single or double-stranded DNA or RNA, including linear or circular single-stranded DNA. Viruses were added at a very low concentration most often corresponding to 3 or 0.8 times the validated level of detection of quantitative reverse transcriptase PCRs (RT-PCRs). For the viruses represented, or resembling those represented, in public nucleotide sequence databases, we show that the higher output of Illumina is associated with a much greater sensitivity, approaching that of optimized quantitative (RT-)PCRs. In this blind study, identification of viruses was achieved without incorrect identification. Nevertheless, at these low concentrations, the number of reads generated by the Illumina platform was too small to facilitate assembly of contigs without the use of a reference sequence, thus precluding detection of unknown viruses. When the virus load was sufficiently high, de novo assembly permitted the generation of long contigs corresponding to nearly full-length genomes and thus should facilitate the identification of novel viruses.

Reannotation of the genome sequence of Clostridium difficile strain 630.

A regular update of genome annotations is a prerequisite step to help maintain the accuracy and relevance of the information they contain. Five years after the first publication of the complete genome sequence of Clostridium difficile strain 630, we manually reannotated each of the coding sequences (CDSs), using a high-level annotation platform. The functions of more than 500 genes annotated previously with putative functions were reannotated based on updated sequence similarities to proteins whose functions have been recently identified by experimental data from the literature. We also modified 222 CDS starts, detected 127 new CDSs and added the enzyme commission numbers, which were not supplied in the original annotation. In addition, an intensive project was undertaken to standardize the names of genes and gene products and thus harmonize as much as possible with the HAMAP project. The reannotation is stored in a relational database that will be available on the MicroScope web-based platform (https://www.genoscope.cns.fr/agc/microscope/mage/viewer.php?S_id=752&wwwpkgdb=a78e3466ad5db29aa8fe49e8812de8a7). The original submission stored in the (International Nucleotide Sequence Database Collaboration) INSDC nucleotide sequence databases was also updated.

In silico comparison of Yersinia pestis and Yersinia pseudotuberculosis transcriptomes reveals a higher expression level of crucial virulence determinants in the plague bacillus.

Although Yersinia pestis and Yersinia pseudotuberculosis are genetically very similar (97% nucleotide sequence identity for most of the chromosomal genes), they exhibit very different patterns of infection. Y. pestis causes plague which is usually fatal in the absence of treatment, whereas Y. pseudotuberculosis generally triggers non-life-threatening intestinal symptoms. This drastic difference in pathogenicity may result from the acquisition of a few species-specific genes, but also from differences in their transcriptional regulation networks. In this study, we performed an in silico comparative whole-genome transcriptome analysis of Y. pestis and Y. pseudotuberculosis grown in parallel under 8 distinct conditions to determine whether they exhibit differences in their regulatory networks. In this analysis, 304 genes common to both species were found to display significant inter-species differences in transcriptional levels, with 91% of them being more expressed in Y. pestis. Remarkably, 3 major virulence determinants conserved in the 2 species (the pYV virulence plasmid, the High Pathogenicity Island, and the ail locus) were among the genes more expressed in Y. pestis. Furthermore, the induction at 37°C of pYV-borne genes was considerably greater in Y. pestis than in Y. pseudotuberculosis. Conversely, the rovA transcriptional regulator gene was more transcribed in Y. pseudotuberculosis. We also performed a clustering analysis of the transcriptome data of both Y. pestis and Y. pseudotuberculosis, which allowed to group genes according to their expression profiles. This analysis identified groups of genes with unknown functions which, based on regulation patterns similar to those of known virulence genes, are potential new virulence determinants in Y. pestis. In conclusion, this is the first comparative analysis at the whole-genome level of the transcription profiles of Y. pestis and Y. pseudotuberculosis. Our results suggest that the higher pathogenicity of the plague bacillus may not only result from the acquisition of new genetic material, but also from a higher expression level of common crucial virulence genes. This in silico analysis thus opens new avenues for investigating Y. pestis gain of pathogenicity and new potential virulence factors.

Genoscape: a Cytoscape plug-in to automate the retrieval and integration of gene expression data and molecular networks.

UNLABELLED: Genoscape is an open-source Cytoscape plug-in that visually integrates gene expression data sets from GenoScript, a transcriptomic database, and KEGG pathways into Cytoscape networks. The generated visualisation highlights gene expression changes and their statistical significance. The plug-in also allows one to browse GenoScript or import transcriptomic data from other sources through tab-separated text files. Genoscape has been successfully used by researchers to investigate the results of gene expression profiling experiments. AVAILABILITY: Genoscape is an open-source software freely available from the Genoscape webpage (http://www.pasteur.fr/recherche/unites/Gim/genoscape/). Installation instructions and tutorial can also be found at this URL.

From a consortium sequence to a unified sequence: the Bacillus subtilis 168 reference genome a decade later.

Comparative genomics is the cornerstone of identification of gene functions. The immense number of living organisms precludes experimental identification of functions except in a handful of model organisms. The bacterial domain is split into large branches, among which the Firmicutes occupy a considerable space. Bacillus subtilis has been the model of Firmicutes for decades and its genome has been a reference for more than 10 years. Sequencing the genome involved more than 30 laboratories, with different expertises, in a attempt to make the most of the experimental information that could be associated with the sequence. This had the expected drawback that the sequencing expertise was quite varied among the groups involved, especially at a time when sequencing genomes was extremely hard work. The recent development of very efficient, fast and accurate sequencing techniques, in parallel with the development of high-level annotation platforms, motivated the present resequencing work. The updated sequence has been reannotated in agreement with the UniProt protein knowledge base, keeping in perspective the split between the paleome (genes necessary for sustaining and perpetuating life) and the cenome (genes required for occupation of a niche, suggesting here that B. subtilis is an epiphyte). This should permit investigators to make reliable inferences to prepare validation experiments in a variety of domains of bacterial growth and development as well as build up accurate phylogenies.

CandidaDB: a multi-genome database for Candida species and related Saccharomycotina.

CandidaDB (http://genodb.pasteur.fr/CandidaDB) was established in 2002 to provide the first genomic database for the human fungal pathogen Candida albicans. The availability of an increasing number of fully or partially completed genome sequences of related fungal species has opened the path for comparative genomics and prompted us to migrate CandidaDB into a multi-genome database. The new version of CandidaDB houses the latest versions of the genomes of C. albicans strains SC5314 and WO-1 along with six genome sequences from species closely related to C. albicans that all belong to the CTG clade of Saccharomycotina-Candida tropicalis, Candida (Clavispora) lusitaniae, Candida (Pichia) guillermondii, Lodderomyces elongisporus, Debaryomyces hansenii, Pichia stipitis-and the reference Saccharomyces cerevisiae genome. CandidaDB includes sequences coding for 54 170 proteins with annotations collected from other databases, enriched with illustrations of structural features and functional domains and data of comparative analyses. In order to take advantage of the integration of multiple genomes in a unique database, new tools using pre-calculated or user-defined comparisons have been implemented that allow rapid access to comparative analysis at the genomic scale.

GenoList: an integrated environment for comparative analysis of microbial genomes.

The multitude of bacterial genome sequences being determined has generated new requirements regarding the development of databases and graphical interfaces: these are needed to organize and retrieve biological information from the comparison of large sets of genomes. GenoList (http://genolist.pasteur.fr/GenoList) is an integrated environment dedicated to querying and analyzing genome data from bacterial species. GenoList inherits from the SubtiList database and web server, the reference data resource for the Bacillus subtilis genome. The data model was extended to hold information about relationships between genomes (e.g. protein families). The web user interface was designed to primarily take into account biologists' needs and modes of operation. Along with standard query and browsing capabilities, comparative genomics facilities are available, including subtractive proteome analysis. One key feature is the integration of the many tools accessible in the environment. As an example, it is straightforward to identify the genes that are specific to a group of bacteria, export them as a tab-separated list, get their protein sequences and run a multiple alignment on a subset of these sequences.

Annotation, comparison and databases for hundreds of bacterial genomes.

The multitude of bacterial genome sequences being determined has opened up a new field of research, that of comparative genomics. One role of bioinformatics is to assist biologists in the extraction of biological knowledge from this data flood. Software designed for the analysis and functional annotation of a single genome have, in consequence, evolved towards comparative genomics tools, bringing together the information contained in numerous genomes simultaneously. This paper reviews advances in the development of bacterial annotation and comparative analysis tools, and progress in the design of novel database structures for the integration of heterogeneous biological information.