gEVAL — a web-based browser for evaluating genome assemblies

1Reference genomes are the foundation for genomic biology. As more and more de novo sequencing projects are being conducted, and more draft genomes released, the continued challenge is to create sufficiently complete and correct assemblies that can be confidently used as references by the research community. Although it has been over a decade since it was announced that the Human genome was completed, this statement referred to the then achievable quality of sequence resolution. There remained errors and lack of sequence coverage for important regions, for example many areas encoding multi-gene families. Since then, genome assemblies of varying quality have been produced for many more species and individuals, often resulting in annotation errors with negative influence on their research use. Groups such as the Genome Reference Consortium (GRC, genomereference.org) lead efforts to curate and resolve genome assembly issues, and maintain the best quality genome references possible for key research species.

2Depending on the project, groups that tackle creating a genome reference may have at their disposal multiple alternative assemblies together with resources such as clone libraries, collections of short reads, cDNA sequences, RNAseq data, physical maps, optical maps or genetic markers. These datasets help in repairing and reorganizing genomic regions or to create a strategy to infuse new sequence into a draft assembly. In parallel, research biologists who observe discrepancies between their data and the genome reference would benefit from being able to assess evidence supporting the reference in the region.

3Current public genome browsers include many of these types of datasets in their analysis repertoire, but usually focus on gene annotation. They also represent major assembly builds released every few years and neither reflect uncertainties in the sequence itself, nor more recent improvements. Here we introduce the gEVAL Browser project, a collection of software and frequently updated databases for key species that takes a tiling path as the backbone, conducts analyses using new sources of data and regularly releases the results in a web interface for users to evaluate sequence integrity and create strategies for sequence management.

4The web based gEVAL browser and the underlying back end databases use the Ensembl project as a framework to build on. Supporting the GRC, it features the current, previous and in-progress official reference genomes of human, mouse chicken and zebrafish while also hosting assemblies of these species not found in other public genome browsers including those curated by the Genome in a Bottle Consortium (Supplementary Table S1), and 16 lab and wild derived mouse strains from the international Mouse Genomes Project. Other species present but less frequently updated include pig (Sus scrofa), rat (Rattus norvegicus) and three helminth species (Echinococcus multilocularis, Strongyloides ratti, Schistosoma mansoni) (Supplementary Table S1).

5There are chromosomal, regional overview, detailed region and comparative browser views, in addition to lists of anomalies. Individual analysis tracks displaying a wide variety of glyphs can be selected interactively. Common tracks include clone library end sequences (Fig. 1A), cDNA alignments (Fig. 1C), genetic markers and self-alignments within the genome. Additional tracks available for some assemblies include issues in regions under review by the GRC, exact clone or sequence component placements and quality of overlaps between neighbouring assembly components (Fig. 1B). The glyphs are coloured to indicate discrepancies and offer pop-up menus that reveal further information and navigational options. All tracks incorporate navigational links to access adjacent s regions, allowing the user to ‘walk’ along the genome from one feature/issue to the next. Custom external user-sourced tracks can also be added by attaching datasets stored in trackhubs or genome browser supported file formats (GFF, BED).

6Comparative genome analysis is invaluable for revealing conserved regions between organisms. Uniquely in gEVAL, comparative analysis focuses on genomic alignments among the different assemblies available for the same species (Fig. 1E). This is useful in capturing sequence differences caused by both variation as well as misassembly, and aids the improvement of one assembly with components/guidance from another (Supplementary Figure S1).

7Unique to gEVAL, it integrates general annotation with long range information produced from recent technologies such as single molecule genome/optical maps (;; Supplementary Table S2). These maps prove invaluable for scaffolding assemblies but are also useful for capturing genome-wide structural variation (Fig. 1D). In addition to presenting evidence, gEVAL can suggest specific sequences from an aligned resource to fill reference gaps, or can suggest inversions or other rearrangements such as expansions or contractions of tandem repeats.

8gEVAL is a web-based browser that allows easy detailed evaluation of genome assemblies through its tools and pre-computed analyses, and suggestion of fixes. It is a key tool used by the GRC for the improvement of the primary vertebrate reference genomes. It is widely used externally for evaluating detailed regions of these genomes in different assembly versions, and also by other reference assembly projects, e.g. the respective international genome consortiums of pig, chicken and mouse. To enable this use the gEVAL code can be downloaded (see Availability and Implementation Section) and installed alongside an Ensembl installation for any other genome assembly.