RefSeq and LocusLink: NCBI gene-centered resources

LocusLink

The locus-to-sequence association maintained in LocusLink is used to provide reciprocal connections among LocusLink, RefSeq and several content-rich resources at NCBI including the Conserved Domain Database (CDD), GenBank (9), Genes and Disease, HomoloGene, Map Viewer, OMIM, PubMed, dbSNP (10), dbSTS and UniGene (11) (Table 2). In addition, nucleotide and protein sequences and OMIM records are linked to LocusLink reports via the NCBI LinkOut service. Following ‘LinkOut’ at the upper right side of an Entrez record is a convenient method to access a wealth of information about that gene. Links between LocusLink and Genes and Disease, HomoloGene, RefSeq, dbSNP and UniGene are based on shared identifiers stored by these resources (e.g., MIM numbers or GenBank accessions). Links to dbSTS and CDD are computed based on electronic PCR (ePCR) (12) and BLAST, respectively. And links to the NCBI Map Viewer are based on sequence identity, cytogenetic location or genetic map data. URLs to these resources are provided in Table 3.

Several organisms and features have been added to the LocusLink web site over the past year and are listed below. LocusLink provides links to each resource whenever additional information is available.

Conserved domains: CDD domains annotated on RefSeq proteins are summarized in the product section of LocusLink reports. This section also includes the protein name, alternate names and enzyme commission (EC) numbers.

GeneRIF: LocusLink now supports community-generated functional annotation. LocusLink reports display submitted GeneRIFs (Gene References Into Function) and include a link to the submission form. Submissions consist of a publication (PubMed ID) and brief comment summarizing the functional information supported by the publication.

HomoloGene: HomoloGene reports UniGene clusters that are either calculated to be homologous, based on sequence identity, or reported to be homologous. HomoloGene also reports the ‘best’ homologous sequence pairs. HomoloGene includes all genomes in UniGene (human, mouse, rat and zebrafish), as well as Drosophila melanogaster mRNAs.

Maps: LocusLink now provides links to the Map Viewer and the Human/Mouse Homology map, as appropriate. The Map Viewer presents graphical views of various maps at the whole genome and single chromosome levels. This resource is currently available for D.melanogaster, Homo sapiens and Mus musculus. Views include presentation of genes, STS markers, SNPs and disease phenotypes along the chromosomes. The Human/Mouse Homology maps display genes, thought to be orthologous, ordered by MGD’s genetic map and the human sequence map.

Organisms: LocusLink now includes fruit fly, human, mouse, rat and zebrafish genes. The query interface supports retrieving results for all species, restricting to a single organism and displaying query results as a ‘Brief’ or expanded ‘Summary’ display.

Reference sequences: The RefSeq section of the LocusLink report has been expanded to include mitochondrial records, genomic contigs and modeled transcripts and proteins produced by the NCBI Human Genome Annotation effort. RefSeq data for a locus may include curated RefSeq records representing known transcripts, a genomic contig and modeled transcripts that overlap, but are distinct from, the known transcripts. Table 1 indicates the different accession number formats and categories they represent.

RefSeq

RefSeq transcript and corresponding protein records are now produced through two independent processes. The first process, which relies heavily on manual curation, provides a non-redundant dataset of transcripts and proteins of known genes for human, mouse and rat; these ‘known’ genes have some supporting evidence for the existence of the gene although the protein function may not be clear. The second process is NCBI’s Human Genome Annotation project, which will be described in detail elsewhere. Briefly, working draft and finished human genome sequence is assembled into contigs (13); gene predictions methods are used to produce the modeled mRNA and protein RefSeq records (see Table 1). The modeled transcripts and proteins may have different degrees of mRNA or EST evidence, or may be predicted ab initio.

RefSeq records include links to LocusLink where additional information is often available. The organisms and features added to RefSeq records over the past year are summarized below.

Category: RefSeq mRNA and protein records are now provided in four categories (previously termed ‘Status’ in LocusLink reports): (i) genome annotation, (ii) predicted, (iii) provisional and (iv) reviewed. The genome annotation category includes contigs, modeled mRNAs and corresponding modeled proteins. Predicted records represent genes of unknown function that are supported by mRNA sequence generated in a full-length insert cDNA sequencing project, homology or ESTs; the location of the protein product may be predicted in these records. Provisional and reviewed RefSeq records largely represent genes with known or inferred function. Provisional RefSeqs are not yet reviewed, whereas reviewed RefSeq records have been individually reviewed by NCBI staff and are richly annotated. Reviewed records include publications, a gene description and annotation that may not be on the provisional RefSeq or original GenBank records. Furthermore the sequence may be modified, relative to the source GenBank sequence, to extend the UTR, provide a transcript variant or remove contaminating vector or linker sequence. By including sequence data derived from more than one GenBank record and from the literature, reviewed RefSeq records may provide a full-length mRNA sequence that is not available in a single GenBank record.

Conserved domains: BLAST analysis is used to compare RefSeq proteins to the NCBI CDD; significant domain matches are annotated on the protein record and are linked to the CDD resource where additional information is available. Reviewed RefSeq proteins may include additional domain annotation derived from the literature.

Links: RefSeq records include links that facilitate access to additional information. Links are provided to: (i) the record(s) from which the mRNA (and hence protein) was derived, (ii) LocusLink, (iii) OMIM, (iv) MGD, (v) RGD and (vi) CDD, as appropriate. In addition, an overview of sequences that share sequence identity is available through the ‘Related sequences’ link at the top right of RefSeq (and GenBank) records.

Organisms: The mRNA/protein RefSeq resource now provides sequence standards for human, mouse and rat. Review of all three organisms to verify the initial ‘gene name-to-sequence’ association and to provide highly annotated reviewed RefSeq records is ongoing.

Variation: Variation features are automatically added to Genome Annotation RefSeq records using the data available in dbSNP. This expanded annotation will also be added to RefSeq records representing known genes.