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

Over the past five years, developments in the technology for determining DNA sequences have led to a substantial change in genome sequencing strategies. Applications in microbial genomics range from re-sequencing collections of strains to non-coding RNA identification, transcription start site determination, and extended metagenome studies [1]. Current high-throughput sequencing devices — a.k.a. “Next-Generation Sequencing” (NGS) technology — can yield hundreds of gigabases of sequence data rapidly and at an ever-decreasing cost. However, the ratio between the production of these sequence data and computer storage capacities is steadily increasing, as is the cost of data management and analysis [2]. As a consequence, the bottleneck affecting large-scale genomics projects is shifting from data generation to data interpretation. Downstream from the primary and secondary levels of analysis — such as quality control, read mapping and de novo assembly — the most time-consuming and specialised tool-demanding tasks are now related to data mining and exploration. To facilitate this analysis step, flexible visualisation interfaces are required [3].

Comparative genomics stands out as one of the major applications of high-throughput sequencing. In the field of microbiology, this discipline involves the elucidation of genetic traits underlying specific (non-)pathogenic phenotypes, and contributes to deciphering evolutionary relationships within large collections of isolates, with direct clinical applications [4, 5]. The most frequent strategy for comparing genome organisations is to examine synteny relationships between genomes, which is the association of gene similarity with gene ordering along the chromosomes. These data can be usefully complemented by information about polymorphism at the nucleotide level; obtained at the genome scale by NGS analyses, biodiversity data help to reveal global trends in evolutionary relationships, and structural variation events such as recombination or insertion/deletion of genomic islands.

A large number of software systems have been developed in the last decade to facilitate comparative genomic studies; they give access to diverse features, in terms of type and quantity of data, analysis methods and styles of user interfaces [6-29] (and see the Discussion below). Many are web-based applications, which tends to hinder interactivity as standard HTML does not enable a dynamic user experience. Some of these tools are more user-friendly, either because they use advanced web protocols (e.g., HTML5) or because they are based on native desktop frameworks; however, these applications are mostly limited to the visualisation of one single or few genomes and/or to limited and particular types of data representation.

Here, we present a novel software application — SynTView — providing the user with substantial flexibility for comparative analysis of microbial genome information. Its originality stems from the combination of many features individually available in other applications: it is a web-based and desktop-capable Flash software; it offers highly interactive graphical user interfaces; it is designed for the visualisation, through many complementary perspectives, of both synteny relationships and DNA polymorphism data for multiple bacterial genomes. The SynTView software is therefore ideally suited to the exploration of data obtained in NGS-based population genomic studies.

The SynTView software was written in ActionScript 3 (AS3) language, using the Apache Software Foundation Flex framework [30] and the associated plug-in for the Eclipse™ Integrated Development Environment (IDE). This technology allows Adobe® Flash software to be built, providing dynamic presentations of complex data through either web browsers or desktop clients. Some interactive data visualisations were implemented using the Flare library [31].

SynTView was built around the following rationale: n genomes should be compared with one reference genome (n can be as large as the computer memory will allow). Depending on the mode of access to the application (see below), genome data can be loaded either by the user using import functions (desktop client) or with the help of a software administrator (web configuration — the only requirement is this case is to place appropriately formatted flat files in the right directory). These data can include complete annotated genomes and associated comparison files for synteny visualisation, or polymorphism data from NGS data sets. Currently, the accepted file formats are ptt (GenBank) for genome annotation, tab-delimited files for other information (protein correspondence, Single Nucleotide Polymorphisms — SNPs) and the Newick format for phylogenetic trees.

The user can access the application in three different ways. First, a stand-alone client can be run using the Adobe desktop application cross-platform runtime AIR (Adobe Integrated Runtime), which will be installed automatically along with SynTView, if required. This allows the users to work with their own locally stored data. Alternatively, web access for multiple users can be implemented; in this configuration, SynTView uses flat files stored on the web server. Finally, SynTView is available in combination with the GenoList integrated environment [32], a multi-genome microbial database accessible on the web, which houses 750 bacterial genomes [33]. This option can exploit the comparative genome data pre-computed in GenoList.

Extensive performance tests have shown that the application reached technological limits if more than ten thousand dynamic graphical objects (i.e., selectable and scalable objects) were used. Drawing strategies were thus developed to ensure that graphical outputs remain below this upper limit, using a mix of static and dynamic objects. For instance, SNP local maps come up with three display modes according to the density of polymorphic positions and the representation scale: (i) in the local view, SNP graphical objects remain interactive (i.e., presence of associated contextual menus, animation of the related gene, etc.) when the genome region displayed is not longer than 200 kb; (ii) if the region is longer than 200 kb, a static drawing is computed; and (iii) SNPs can also be represented as blocks that cluster a series of adjacent polymorphic positions together (the number of SNPs per kb is indicated by the colour of the block) — these blocks are clickable to return to the interactive local view.

The software features described in this paper illustrate only some of the main capabilities of the SynTView application. A full list of its functionalities is available in Additional file 1.

One of the main strengths of SynTView lies in its highly interactive graphical user interface, possible due to the Flash technology used to develop the software. This technology is increasingly exploited by comparative genomics applications [21, 36-38]. It allows the development of rich Internet (or client) applications, which escape the limitations faced by standard HTML websites. Flash also provides advantages similar to or greater than those of other technologies (e.g., JavaScript and SVG): the Flash plug-in is installed out-of-the-box on most computers and it is fully portable across different web browsers. Although HTML5 is becoming more and more popular on the world-wide web, including for genomics applications (e.g., [39]), it is currently less cross-compatible among web browsers than Flash and mostly limited to playing videos and animations, rather than rich Internet applications; indeed, development environments and resources are still far more efficient when using Flash/Flex (e.g., the object serialisation protocol AMF3 in ActionScript/MXML) than using HTML5. In addition, the Flash strategy allows us to propose a web access and a client application (AIR environment) simultaneously. Thus, we have created a versatile solution that offers the best of both worlds: a desktop program that can easily be installed by a non-expert user, providing direct access to his local storage and file systems, without needing an IT infrastructure to deploy complex web architectures; but also, a browser application that may be more intuitive for some users and allows several people to access a single instance of the software, while retaining many of the characteristics of desktop-like programs.

Note that SynTView is not intended for data generation: the software is focused on the dynamic exploration of genome data produced externally. In particular, SynTView is not linked to any particular database, which allows the generation of plain text files that the application will interpret and subsequently represent (thus easily integrating the user’s own data) or the exploitation of other sources of information, such as the GenoList environment.

SynTView was designed to be extremely fast, and sports a user-friendly interface that is straightforward to use for biologists accustomed to graphical widgets. In particular, data exploration is facilitated by the possibility of connecting several views: indeed, in both the synteny and SNP visualisations, the various representations are interactively linked together, and a user selection (e.g., genes, genome region) in one panel dynamically triggers either a related selection or a view modification in other panels (Figure 1). This is a major feature that is very helpful for making sense out of large and complex data sets, in a way that is not available in most similar software, even if individual analogous views can be found (see below). For example, interesting areas of the chromosome may be observed (e.g., recombination, inversion) using the line plot representation; it is possible to zoom in on these regions, then select the corresponding BDBH to export them, or shift with one mouse click to the local view to analyse the region in more detail. The export functionalities offered by SynTView are also appropriate for further analysis, once an initial exploration has identified the data sets to be selected. The images generated by the software can be saved in PNG format and subsequently used in reports or publications.

Numerous applications are already available for visualising comparative genomics data from microbial species, but we believe that SynTView fills a gap and is more complete: it is a flexible software package, compatible with various computing environments, and uniquely integrates complementary features present, in isolation, in similar tools. For instance, the CGAT Java alignment viewer [28] and the GSV web package [23] are efficient for syntenic visualisations, but are restricted to pairwise genome comparisons. Likewise, MizBee [19] implements an innovative multi-scale synteny browser that enables interactive exploration of various kinds of relationships, but only between two multi-chromosome genomes at a time. The Cinteny [25] and Genomicus [18] web sites are primarily designed for metazoa; Genomicus focuses on ancestral genome reconstruction and Cinteny provides on-the-fly computed, but non-interactive, synteny views. CGView [14] and Easyfig [26] are highly customisable packages for producing publication-quality, but static-only, figures; the GView [22] Java application is more interactive but mainly focused on the display style configuration. Companion tools to the popular genome browser GBrowse, such as SynBrowse [20] and SynView [29], also offer interactivity albeit limited. Circos [17] and BRIG [7], available as desktop applications, are mostly restricted to static (circular) views. GenomeRing [15] has an interactive web interface for visualising multiple circular genome alignments, based on an original concept of a common annotation coordinate system. Gobe [21] is a web-based Flash application, hence very dynamic, but it is limited to synteny information. Although the Mauve [11] and VISTA [13] packages come with useful Java visualisation interfaces, they are primarily dedicated to whole genome alignments. The MGV [16] and Sybil [24] suites are feature-rich web-only applications for the interactive visualisation of microbial genomic contexts, but these software do not include nucleotide polymorphism information. More recent software were developed especially to handle NGS data sets, although most do not focus on the representation of processed biodiversity information for large collections of microbial strains. They include: IGV [27] and Savant [12] which are both widely used; and GenomeView [6] which is a feature-rich stand-alone genome browser able to cope with dozens of “NGS” genomes, but which lacks integrated circular representations. The latest version of Artemis [8], an established annotation tool, in association with ACT [10] and DNAPlotter [9], provides the user with a large set of comparative views, some of them being dedicated to bacterial population genomics studies. Together, these three Java applications probably come closest to matching the features provided by SynTView.

SynTView has been used in several genome sequencing projects. Its data visualisation features help to reveal or confirm biological models. For instance, it allowed us to reproduce an observation regarding bacterial genome organisation: the location of the superintegron in Vibrio species [40] (Figure 2). SynTView was also used to analyse hundreds of bacterial strains that were Illumina-sequenced and whose nucleotide polymorphism was determined, thus yielding SNP heat map representations similar to those in [41] (Figure 3). Other applications include the analysis of genomic islands visualised by the sequencing coverage representation, allowing confirmation, or otherwise, of the presence or absence of such islands, and the identification of specific deletions therein.

SynTView will be further developed in several ways. First, we plan to increase the scope of the software by accommodating other types of data. In particular, SynTView should enable the management of incomplete genomes for synteny representations. For NGS, we also intend to design novel views for RNA-seq data (such as expression levels or transcription start sites) and to improve the SNP representation such that it can handle multiploid organisms. As far as end-user interactions are concerned, improvements will include features such as the ability to save selected views and gene baskets in a permanent way.

We have developed software for the dynamic visualisation of microbial genomes, such that both synteny and DNA polymorphism can be displayed. The most important asset of SynTView is its interactivity, inherent to the use of the Flash technology. In addition, several properties of the application facilitate efficient investigation of large data sets at the chromosome scale: for instance, its flexibility for establishing links with various user data sets and the dynamic interactions between the numerous specialised views. This innovative approach to data exploration is not possible with most existing applications, because they are more static and/or do not offer multiple views of multiple genomes.

SynTView is available under the GNU LGPL (Lesser General Public License) 3.0 license at the following URL: <http://genopole.pasteur.fr/SynTView>. As the Flash runtime is pre-installed in most web browsers, the application can run without any further installation. Technical documentation, tutorials and demonstration sites are available at the following URL: <http://genopole.pasteur.fr/SynTView/documentation>.

Project name: SynTView

Project home page: http://genopole.pasteur.fr/SynTView

Operating system(s): Platform independent

Programming language: ActionScript 3

Other requirements: Adobe® Flash Player

License: GNU LGPL 3.0

Any restrictions to use by non-academics: none