An Automated Genome Annotation Pipeline for the i5K Pilot at the Baylor College of Medicine Human Genome Sequencing Center

The i5K is an initiative to sequence the genomes of 5,000 arthropods of medical, agricultural and scientific importance. As a pilot project to identify potential problems and challenges in species selection, identification and acquisition, DNA isolation, sequencing strategies and assembly, automated and manual annotation, analysis and publication, the BCM-HGSC is sequencing ~30 arthropods, selected by the i5K species selection committee.

We present a new automated annotation pipeline for arthropods that enables the automated annotation of high quality gene models for 1,000’s of insect genomes. The annotation pipeline is based upon the combination of a modified version of the CEGMA pipeline adapted to use profile HMMs for several thousand single-copy arthropod genes and the Maker pipeline using extensive metazoan protein alignments, species-specific RNAseq evidence as well as automatic training of the ab initio Augustus and SNAP gene predictors. The output of the automated pipeline has been validated against a set of ~3,000 ortholog genes, identified in the assembly and has > 98% specificity and sensitivity at the base pair level for genes present in the assembled genome sequence. Genes are given provisional names based upon a heuristic combination of orthology predictions and reciprocal best BLAST hits with Drosophila, C. elegans, and Human. Approximately, 30 evidence tracks with evidence partitioned by phyla are generated for the web-Apollo tool, enabling manual annotators to see all input evidence for a gene model. To date, 20 species have been put through the i5K pilot annotation pipeline, and are being hosted by the National Agricultural Library providing access to search, browse, and for community manual annotation.

This pipeline is one of the tools that is necessary for the maximum utility of the thousands of arthropod genomes envisioned by the i5K genome initiative. Our future aim is to port this pipeline to the cloud, enabling facile use by any researcher, without installation headaches, and to enable hundreds of genomes to be automatically annotated in parallel. We hope this will be one of the tools bringing the genomics revolution to all the arthropod species.