GOA December 2009
Gene Ontology Annotation at UniProtKB, 2009
Report on the GOA team's activities between March 2009 and September 2009.
Swiss-Prot contributors (EBI, Hinxton, UK and SIB, Geneva, Switzerland):
Ioannis Xenarios Amos Bairoch Lydie Bougueleret Serenella Ferro-Rojas
Andrea Auchinchloss Marie-Claude Blatter Emmanuel Boutet Lionel Breuza Alan Bridge Paul Browne Wei Mun Chan Elizabeth Coudert Louise Daugherty Ruth Eberhardt Anne Estreicher Livia Famiglietti Marc Feuermann Rebecca Foulger Nadine Gruaz-Gumowski Ursula Hinz Silvia Jimenez Florence Jungo Guillaume Keller Kati Laiho Duncan Legge Philippe Lemercier Damien Lieberherr Michele Magrane Ivo Pedruzzi Sylvain PouxCatherine Rivoire Bernd Roechert Michel Schneider Eleanor Stanley Andre Stutz Shyamala Sundaram Michael Tognolli
We continue to put emphasis on the annotation of those genes selected for the Reference Genome Project.
In addition, with the newly started kidney-centric annotation project, additional emphasis has been placed on certain genes associated with renal development and disease.
Currently the curators from the GOA and BHF-UCL projects have together completely annotated 77.5% (609/785) of supplied Reference Genome Targets.
Between January 2009 and November 2009, the GOA project provided the GO Consortium with ten annotation file releases, including non-redundant sets of GO annotations to the human, mouse, rat, zebrafish, arabidopsis, chicken and cow proteomes, as well as data releases for annotations of all proteins in UniProtKB. GOA now provides over 54 million GO annotations for 6.7 million proteins in over 205,000 different taxonomic groups. GOA provides 167,781 annotations for the human proteome (providing 91% of the human proteome with at least one GO annotation). Over the last year the number of manual annotations has increased by 5.4 % in the UniProtKB file and the number of manual annotations for the human file has increased by 10.8%. Between January and November 2009, GOA has trained, checked and supported 30 curators in the Swiss-Prot team at the Swiss Institute of Bioinformatics, who have since created over 17,500 manual GO annotations for UniProtKB entries from a range of species.
Methods and strategies for annotation
- Literature curation:
Literature curation continues to be the major focus of our annotation efforts, with an emphasis on the use of experimental evidence codes.
- Computational annotation strategies:
GOA provides IEA annotations from the following methods:
- Swiss-Prot Keyword 2GO (SPKW2GO)1,2
- Swiss-Prot Subcellular Locations2GO (SPSL2GO) 1,2
- Ensembl Compara
1: mapping tables created and maintained by the GOA group
2: electronic annotations generated by the GOA group, using UniProtKB.
Because of the multi-species nature of the UniProtKB, GOA is able to assist in the GO curation (by both manual and electronic means) of proteins from over 205,000 taxonomic groups. However GOA curators are also primarily responsible for supplying the GO Consortium with manual and electronic GO annotations to the human proteome.
GOA curators do not only manually annotate, but also coordinate and check the integration of manual GO annotations from other curation efforts at the EBI (including from InterPro, IntAct and UniProtKB groups).
- Priorities for annotation
- Genes assigned by Reference Genome Project (Rachael, Emily)
- Genes associated with renal processes (Yasmin)
- Requests from user community (all curators)
- Proteins annotated during Swiss-Prot curation duties (all Swiss-Prot/UniProtKB curators at the EBI and SIB)
Presentations and Publications
A. Ontology Development Contributions:
160 SourceForge items regarding requested changes to the GO have been placed by curators associated with the GOA group since March 2009
B. Annotation Outreach and User Advocacy Efforts:
The GOA group has been training Swiss-Prot curators at the Swiss Institute of Bioinformatics in Geneva over the last year. GOA curators have been training small groups of curators, and then evaluating and feeding back to individual curators every GO annotation made during the first two months after training. After this time, if it is felt the curator is confident in annotating to GO, their annotations are subsequently spot-checked. This has been/is a very large amount of work for our group. GOA has traveled to Geneva 5 times since January, training on average 5/6 curators during each visit.
The Swiss-Prot team in Geneva have so far generated approximately 13,158 manual GO annotations to 2,958 UniProtKB proteins.Annotations are created in GOA's protein2go tool, and released in the groups gene association files. Such annotations use the existing source 'UniProtKB' (for column 15 of the gene association file). GOA will continue to train and mentor SIB curators over 2009 and 2010.
Changes to SPKW2GO and SPSL2GO
With the aid of Serenella Ferro-Rojas at SIB, the SPKW2GO and SPSL2GO mappings have been revised - so far 70 mappings have been added to the SPSL2GO mapping, and changes to 24 different SPK2GO mappings have been made since January 2009.
Please note that GOA is intending to stop using the SPWK2GO mapping for production of cellular component GO annotations. In future only the Swiss-Prot Subcellular Location 2 GO (SPSL2GO) mapping will be applied to generate CC annotations from UniProtKB. This change will improve the correctness of GOA-supplied cellular component annotations, as SPSL2GO includes 'host x' subcellular location terms, in contrast to Swiss-Prot keywords. Therefore it is important that all groups which integrate annotations from SPKW2GO from GOA gene association files include annotations originating from SPSL2GO.
Renal GO annotation initiative funded by Kidney Research UK.
This grant started on the 1st April, managed by Yasmin Alam-Faruque, who joined GOA from the Swiss-Prot team at the EBI.
Aim 1: Promote wider scientific engagement in GO annotation through a variety of outreach efforts • Generated web pages providing further information about the Renal GOA Initiative: http://www.ebi.ac.uk/GOA/kidney/ and http://www.geneontology.org/GO.renal.shtml. • Set up the Renal Interest Group mailing list on the Gene Ontology Consortium web pages: http://www.geneontology.org/GO.interests.shtml?all#renal (aimed at encouraging researchers to provide suggestions/ advice/ discussions on renal gene/ protein-related issues) • Two quarterly newsletters have been circulated, which provide further information on the project’s progress: http://www.ebi.ac.uk/GOA/kidney/newsletter/.
Aim 2: Annotate gene products associated with kidney development and disease • In collaboration with members of the Scientific Advisory Panel, a list of renal-related proteins has been created which act as an initial set of curation targets, available from: http://www.ebi.ac.uk/GOA/kidney/. Initial annotation targets were highly investigated proteins involved in the proton pump process and ammonium transport; both processes being important for the maintenance of acid-base homeostasis in the kidney. Additional targets have been provided by the GUDMAP Consortium, which have no current GO annotations. • To date this initiative has created 1,885 GO annotations to 387 UniProtKB protein accessions; this figure exceeds stated annotation targets in the grant proposal • Annotations to this list are publicly available from QuickGO browser (http://www.ebi.ac.uk/QuickGO/GAnnotation?protein=KRUK). - enabling members of the biomedical research community to search and view the dataset easily in the GOA group’s QuickGO browser • Collaboration initiated with the Reactome group, resulting in further members of the solute-carrier transmembrane transporter protein superfamily added to the Reactome database. This is going to be expanded to other transmembrane transporter protein families including the proton pump, ion channels and aquaporins. The inclusion of more proteins involved in renal-specific processes into the Reactome database will provide an even more unique and comprehensively detailed functional dataset for mammalian gene products implicated in renal function and development. • A co-annotation project has been initiated with other curators to improve annotation of proteins involved in the development and function of the Loop-of-Henle structure for mammalian and non-mammalian organisms . This will highlight not only biological insights into the similarities and differences of the orthologous genes in distinct species, but also demonstrate the usefulness of focused, collaborative cross-species GO annotation by demonstrating to users the usefulness of functional annotation. This effort will also lead to focused development of GO terms to more accurately describe renal-associated processes. An envisaged publication on such a study would provide publicity to the renal annotation initiative and other databases.
Aim 3: Improve the Gene Ontology with respect to Renal Processes • Collaboration initiated with the GUDMAP Consortium. This effort is reviewing the state of renal GO terms that currently exist in the ontology (initially with respect to nephrogenesis) and hence develop and improve GO terms in-line with the kidney anatomy work carried out by the renal community
Verification of mappings to UniProtKB accessions in GO Consortium gp2protein files
The GOA group has recently provided groups in the GO Consortium with checks of the UniProtKB accessions applied in gp2protein mapping files. Annotation groups receive an email to indicate where in their file a secondary or deleted UniProtKB has been used. This email will also (where possible) indicate suitable replacement UniProtKB accessions. Such checks will be run and results emailed to annotation groups on the first of each month.
Gene Association File changes
November 2009 As of November, GOA gene association files exclude annotations that apply a secondary or obsolete UniProtKB accession number in either column 2 (DB_Object_ID) or column 8 (With). This has meant that 3,929 (4.5%) manual annotations integrated from external annotation groups have been filtered from GOA files. We are working closely with all collaborating databases to help ensure that their set of UniProtKB accessions are regularly updated to valid primary UniProtKB accessions.
October 2009 gene_association.goa_pdb file needed to be changed in October, as GOA was no longer able to provide annotations to specific PDB chains. Annotations in this file now only associate PDB entries with GO terms. As an example, an annotation line in the PDB file has changed from:
PDB 102M_A q_A GO:0005506 GOA:interpro IEA InterPro:IPR000971 F protein_structure taxon:9755 20090629 UniProtKB to: PDB 102M 102M GO:0005506 GOA:interpro IEA InterPro:IPR000971 F protein_structure taxon:9755 20091001 UniProtKB
GOA, PDBe and InterPro are working together to provide a new file in the 2010, which will once again provide chain identifiers for PDB entries. This new file should hopefully provide an improved set of annotations from InterPro2GO, as well as include manual annotations where the PDB chain is identical to 80% of a UniProtKB accession.
July 2009 Previously, annotation lines produced by electronic methods (as indicated by the presence of the 'IEA' evidence code in column 7) contained two identifiers piped together in the reference column; an internal GOA keyword (e.g. GOA:interpro) and a GO reference identifier (e.g. GO_REF:0000002). As the July 2009 release, the GOA internal reference (e.g. GOA:interpro) and pipe has been removed from this field, so that only a GO_REF identifier is provided.
May 2009 DB field (column 1). Column 1 of the gene association file is used to identify the database which has supplied the sequence identifier displayed in column 2. Its value is very often 'UniProtKB'. However recent changes to other fields in the GOA gene association files have made it difficult for users to identify whether an UniProtKB accession originates from the UniProtKB/Swiss-Prot or UniProtKB/TrEMBL databases. Therefore it is intended that when a UniProtKB accession is provided in column 2, column 1 displays either 'UniProtKB/Swiss-Prot' or 'UniProtKB/TrEMBL'.
February 2009 Human Gene Association File changes. In February 2009, the production of the gene_association.goa_human file changed from using the International Protein Index (IPI) to using the complete human proteome set available from UniProtKB/Swiss-Prot (http://www.uniprot.org/news/2008/09/02/release).
The name and format of this human file has remained the same, however annotations are now assigned to proteins from just the 'UniProtKB' (column 1) database source. Human IPI identifiers continue to be included in column 11 of annotations.
In addition, new releases of the cross-references file for human IPI set (human.xrefs.gz), will no longer be provided. Instead, identifier mapping is possible using the UniProt ID mapping file, available from: ftp://ftp.ebi.ac.uk/pub/databases/uniprot/current_release/knowledgebase/idmapping/idmapping.dat.gz
idmapping.dat.gz is a tab-delimited table, which includes mappings for 20 different sequence identifier types, including IPI identifiers.