An extensible genome annotation workbench based on the Galaxy Platform

Introduction Falling costs of genetic sequencing have allowed sequencing and annotation of the genomes of non-model organism. In annotating non-model genomes, mRNA-seq data has great potential to improve annotation quality. For example the Asian seabass (L. calcarifer) genome annotation effort drew on previously assembled mRNA-seq data provided by the Temasek Life Sciences Laboratory (TLL, Singapore). At the South African National Bioinformatics Institute (SANBI) we undertook gene annotation on the Asian seabass genome using a pipeline built out of custom scripts. Simultaneously, a team at Saint Petersburg State University undertook the same task using MAKER2. Comparing the results of this annotation highlighted the impact of tool and parameter choice in gene prediction. The Galaxy framework allows workflows to be constructed in a high level workflow language that hides the system-specific details of their implementation. We implemented genome annotation workflows in Galaxy, demonstrating its suitability for constructing an annotation workbench  that incorporates re-usable and replaceable modules. Conclusion  Repeatable genome analysis workflows allows for reuse of methods and reproducibility of results. Galaxy workflows allow workflow construction in a flowchart-like idiom similar to the way in which workflows are documented. We demonstrate the construction of two workflows as part of a larger annotation workbench and their use in the annotation of the L. calcarifer genome. Exposing results through Jupyter notebooks and export to browsers such as JBrowse and the (SANBI-authored) Bass Explorer allows results to be examined seamlessly within Galaxy. Future Work We intend to implement further workflows to provide a complete Galaxy-based eukaryotic genome annotation workbench. We will enhance tools that export to an annotation database browser modelled on the Bass Explorer.