Ref Gen pub draft
From GO Public
The Reference Genome Annotation Project
With more and more genomes being sequenced, we are in the middle of an explosion of genomic information. The limited resources available to manually annotate the growing number of sequenced genomes imply that automatic annotation will be the method of choice for many groups. The GO consortium coordinates an effort to maximize and optimize the GO annotation of a large and representative set of key genomes ('reference genomes'). The goal of this project is to completely annotate 12 reference genomes so that those annotations may be used to effectively seed the automatic annotation efforts of other genomes.
The reference genomes are:
- Arabidopsis thaliana (http://www.arabidopsis.org/)
- Caenorhabditis elegans (http://www.wormbase.org/)
- Danio rerio (zebrafish; http://zfin.org)
- Dictyostelium discoideum (http://www.dictybase.org/)
- Drosophila melanogaster (http://flybase.org/)
- Escherichia coli (http://ecolihub.org/)
- Gallus gallus (http://www.agbase.msstate.edu/)
- Homo sapiens (http://www.ebi.ac.uk/GOA/human_release.html)
- Mus musculus (http://www.informatics.jax.org/)
- Rattus norvegicus (http://rgd.mcw.edu/)
- Saccharomyces cerevisiae (http://www.yeastgenome.org/)
- Schizosaccharomyces pombe (http://www.genedb.org/genedb/pombe/)
The Reference Genome GO Annotation Team, with trained and highly skilled GO curators from each genome annotation group, coordinates annotation, facilitates implementation of GO Consortium annotation priorities, and provides quantitative measures to assess progress toward the goal of broad and deep annotation of the reference genomes. This group represents the annotation expertise within the GO consortium and provides key liaisons to the model organism databases that have primary responsibilities for the annotation of the reference genomes.
Priorities for Annotation
Our ultimate goal is to provide comprehensive GO annotation for all gene products in each of the reference genomes. This is a huge task and requires prioritizing curation targets. Our initial annotation efforts (Aug 2006-Sept 2007) focused on orthologs of human disease genes, but in Oct 2007 we widened our list to four priority areas:
- Orthologs of human disease genes.
- Topical or ‘hot’ genes.
- Genes conserved from E. coli to human but currently lacking GO annotation.
- Genes involved in biochemical/signalling pathways.
Each month we curate genes from each category as selected by one of the participating databases on a rotational basis.
Overview of project strategy
Every month each database curates the same set of genes from our priority list. Working on the same genes together promotes cross-organism discussion about annotations and frequently leads to new terms being added to the Gene Ontology. We start from a set of genes selected from the Human Genome.
- Identify the ortholog(s)/homolog(s) of the selected target genes in each species. Not all species may have orthologs/homologs to selected genes.
- Enter the gene identifiers in a shared spreadsheet so that all curators can see the set of genes being curated.
- Collect and annotate available literature about the genes.
- Assign GO terms based on experimental data.
- Review additional (pre)existing GO annotations to make sure they conform to GO annotation standards.
- Record in shared spreadsheet that gene is considered to be comprehensively annotated as of a given date.
A web tool for reference genome annotation is under development. This will help curators to track and compare annotations, thus streamlining the annotation process.
How does this project differ from standard GO annotation?
The reference genome databases have agreed to follow guidelines that are more stringent than those used for standard GO annotation:
- Experimental evidence codes (IDA, IMP, IGI, IPI, IEP) should be used where possible. The ultimate objective would be to provide experimentally-based annotations for all gene products from these organisms.
- Terms inferred from sequence and structural similarity (ISS) should only be used where the terms are supported by experimental evidence for the similar sequence.
- Non-traceable author statements (NAS) should not be used.
- No new annotations should be based on traceable author statements (TAS); existing terms assigned with TAS should gradually be replaced with the appropriate experimental evidence code based on the primary literature.
How do we know when GO annotation is comprehensive?
The amount of literature per gene is very variable. Where possible, we review every paper about a given gene for each organism and capture all possible GO terms. This is only really feasible when there are tens of papers. For genes associated with hundreds or even thousands of publications, we cannot read all of the papers. So we seek to prioritize the literature and capture all functional attributes of the gene in the annotations. In these situations, we often start work with recent reviews that lead us to key experimental papers. Users are encouraged to notify us if we have failed to capture some aspect of a specific gene. Send your comments to GO Help.
When there are no experimental data for any of the reference genome species, but experimental data are available in other model systems, we submit GO annotations for the relevant species to GOA so that this information is captured from the primary literature.
Where can GO annotations from the project be viewed?
It is also possible to specifically view the reference genome effort. A graphical representation of the annotations to a particular gene across the reference genomes can be viewed in full here:
Each curated reference gene links to one graph. In addition to the graph, each page includes two informative tables: a table comparing organism annotations for each term (rows are GO terms, columns correspond to organism), or a table that shows full experimental annotations in each organism for the given gene. This facilitates comparison of the curation status in the 12 reference genomes and helps curators to identify genes that need attention.
This project aims to improve annotations across a wide range of organisms. The resulting high quality annotations will no doubt improve electronic annotations that propagate from this resource and annotations will facilitate cross-species functional comparison. Furthermore, the easy comparison of annotations between organisms may lead to new hypotheses and thus inspire new exciting research.
- Monthly Conference Calls
- First Reference Genome Annotation Meeting, Princeton, NJ, Sept 26, 27, 2007
- Provide coordinated orthology predictions for reference genomes in collaboration with major orthology dataset developers.
- Provide summary metrics on our progress.
- Improve efficiency of the literature annotation process through incorporation of shared annotation tools and information extraction techniques.