Third Party Annotation Database

NCBI and its collaborating databases, DDBJ and EMBL, have established the inferential portion of the Third Party Annotation (TPA) database to accommodate a wider range of submission types. The TPA database was established in 2002 to allow researchers to submit their own analyses of existing GenBank sequences. TPA submissions may include genomic or transcript sequences assembled from primary data, the annotation of features such as genes, coding regions, and transcripts, or functional annotations of protein sequences. The analysis of the primary sequence data will become increasingly important as unannotated data from genome sequencing and EST projects accumulates. Prior to establishing guidelines for the inferential portion of the TPA database, direct experimental evidence was required for TPA submissions.

To enlarge the scope of third party annotations, the new inferential component of the TPA database allows submissions of sequences and features based on analysis of existing GenBank sequences without direct experimental evidence. This original TPA data and all new TPA submissions that include direct experimental support are now included in the experimental section. Inferred TPA submissions, like their experimental counterparts, must be published in a peer-reviewed journal in order to be released.

TPA records can be retrieved or combined with other Entrez queries using the search term 'tpa [Properties]', and the specific experimental or inferential records can be distinguished by the keywords 'TPA:experimental' or 'TPA:inferential'. TPA records are identified in Entrez or BLAST search results by the 'TPA_exp:' or 'TPA_inf:' labels at the beginning of the definition line visible in the example of Fig. 1.

An example of an inferred TPA record is the assembly and annotation of the complete chloroplast genome for the green alga Chlamydomonas reinhardtii (accession BK000554) shown in Fig. 1.

Click on image to view larger

Figure 1. The Chlamydomonas reinhardtii chloroplast genome assembled from 101 other GenBank records that are listed in the 'Primary' field. These individual components can be retrieved by following the 'Components' link in the 'Links' menu. Inferred records are easily recognized by the abbreviation TPA_inf in the DEFINITION or TPA:inferential in the KEYWORD sections of the GenBank record.

In this example, as with all inferrential records, there is indirect experimental evidence for the sequence and new annotation including independent evidence for the individual CDSs, structural RNAs, and other features on the organelle's genome assembled from overlapping primary sequences. Genes or other features predicted by computer programs without any further evidence are not accepted in the TPA database as either experimental or inferential submissions.

For more information on TPA submissions, see:

—MB

www.ncbi.nih.gov/Genbank/TPA.html