The NCBI Taxonomy Project
The purpose of this note is threefold:
(1) to outline the taxonomy project that we have been working on at the
NCBI for the past year,
(2) to solicit volunteers from the taxonomic and phylogenetic communities
(to help curate the taxonomy) and from the users of the sequence
databases (to help in identifying problems with the taxonomy), and
(3) to establish contact with culture collections, stock centers, herbaria
& museums, and any other groups that are maintaining general and/or
specialty taxonomies and/or phylogenies.
The problems with taxonomies used by the sequence databases are well-known:
each of database comes with its own taxonomy; each is different from the
others, and none of them are in full agreement with the current taxonomic
consensus (even if we could imagine that such a thing existed), and all of
them contain a wide variety of different kinds of errors and inconsistencies.
At an even more basic level, it is not always possible (even within the same
database) to determine if two entries come from the same species.
We have developed a taxonomy database management tool (the TaxMan) which is
based on a tree-structured database appliction developer's tool (the TreeTool).
This tool includes a rich set of functions for merging & crossmapping trees.
We have used the TaxMan to build representations of each of the sequence
database taxonomies, as well as a few other taxonomies obtained from other
sources (the ICTV international standard taxonomy for the viruses, the USDA
taxonomy for the plants, and the FlyBase taxonomy for the Drosophilidae).
We have used the TaxMan to merge all of these taxonomies into a single tree,
which we can associate with the database that we are maintaining which merges
all of the sequence databases into a single structure.
After we had merged the sequence database taxonomies, a workshop was
organized by Mitch Sogin, of the Marine Biological Laboratory at Woods Hole
in order to review and revise the taxonomy and to discuss mechanisms by
which the taxonomic community could maintain the taxonomy (as new species
enter the databases and as the taxonomic consensus develops). This workshop
included a dozen representatives, each specializing in different branches
of the taxonomic tree, and included both classical and molecular systematists.
The revised 'backbone' tree will be much more of a phylogenetic taxonomy
than a classical taxonomy; we feel that this will be of more general
use to the users to the molecular sequence databases.
The nucleic acid sequence database collaborators (EMBL and DDBJ) have agreed,
in principle, to adopt the revised taxonomy as a database standard.
We realize that for any given taxonomy there will be at most one person in
the world who is completely happy with it. Although we need a single 'backbone'
tree to associate with the sequence databases (and we will try to make this
tree as good as possible) we do not want to claim that our tree is the
canonical international standard taxonomy. We plan to develop the TaxMan to
make it easy for concerned users to modify the 'backbone' taxonomy as they see
fit, crossmap their personal tree back onto the 'backbone' taxonomy, and index
the sequence databases through their own tree.
For example, we have promoted the Archaea to kingdom level, alongside the
Eubacteria and the Eucaryotae. Others may wish to use the traditional
classification (with the Archaea and the Eubacteria buried in the Procaryotae)
or other modern reclassifications (e.g. the Eocytes). As another example,
we plan to move the birds (Aves) beneath the Archosauria, as a sister group
to the Crocodylia.
As a consequence of this approach, since we are moving towards a phylogenetic
taxonomy instead of a classical taxonomy, the classical concept of taxonomic
rank names (e.g. family, order & etc.) disappears. In the revision of the
protozoan taxonomy which we have recieved, the familiar rank-level suffixes
(-idae, -ida, -iformes, etc.) have been replaced with the generic suffix
(-ids). We will, however, retain the other names (like Kinetoplastida,
Trypanosomatidae & etc.) as synonyms in the tree, so that users may continue
to retrieve the same set of organisms with these names.
There are several consequences for the database users & submitters. First,
we plan to formalize the use of organism names in the database - to collect
all of the variant spellings, synonyms and misspellings and to select a
preferred scientific name for each organism. Second, we plan to phase in
(and retrofit the databases with) new taxonomic classification lines from
the revised tree as the subtrees are returned by the workshop participants.
And finally, each of these fields may change in new releases of existing
entries in the databases, as new synonyms and misspellings are identified in
organism names and as the taxonomy is revised to reflect new work in the field.
Submitters who wish to associate different names and taxonomic classification
lines with their entries will be allowed to enter this information in a /note
attached to the source feature in the flatfile format.
We have added a directory to our anonymous ftp site (ncbi.nlm.nih.gov,
a.k.a. 18.104.22.168) to make available files associated with this project.
In the directory "repository/taxonomies/taxman" you will find:
id (15Mb) - the ASN.1 text formatted version of the merged taxonomy
id.bin (7Mb) - the ASN.1 binary formatted version of the merged taxonomy
id.report.ps - a text-file report of the merged taxonomy (375 pages)
id.report.index.ps - a text-file index to accompany id.report (139 pages)
manual.ps - the first half of a user's manual for the taxman
taxman - the Sun executable file for the taxman program
Please send comments, criticisms & suggestions to federhen at ncbi.nlm.nih.gov
It will be a long project to clean up the taxonomy and to retrofit the
sequence databases, but we hope that with the help of the several communities
involved, we will be able to add a very powerful, uniform & useful set of
tools for retrieving and manipulating the information in the sequence
Scott federhen at ncbi.nlm.nih.gov