MATERIALS AND METHODS
Biologists from University of Missouri at St. Louis and Missouri Botanical Gardens, and curators from the TAIR, MaizeGDB, and Gramene databases worked together to evaluate growth and development in Arabidopsis (Arabidopsis thaliana), maize (Zea mays), and rice (Oryza sativa), examining the vocabularies and models used to describe the whole-plant growth stages in each species. Growth stages of Arabidopsis were described by Boyes et al. (2001) based on the BBCH scale (Meier, 1997) that includes both monocot and nonmonocot species. The BBCH scale in turn is based on the Zadok scale, developed for Triticeae (Zadok et al., 1974), which forms one of the literature bases for the cereal GSO developed by Gramene database (Jaiswal et al., 2006). Rice terminology was derived from INGER (1996), for Triticeae from Zadok et al. (1974) and Haun (1973), and for sorghum from Doggett (1988). MaizeGDB (Lawrence et al., 2005) derives its growth stage vocabulary from a modified version of Ritchie's scale (Ritchie et al., 1993). The vocabulary developed by MaizeGDB was integrated into cereal GSO in the Gramene database as well. With these preexisting interconnections in the core databases, we were able to begin synthesizing them into a generic ontology. Similar growth stage concepts for the above species were identified and mapped to the generic growth stages and stored in mapping files. The mapping files are available at http://brebiou.cshl.edu/viewcvs/Poc/mapping2po/. More details about the project and ontology development is available on the documentation section of the PO Web site (http://www.plantontology.org/docs/docs.html).
Review of Ontology
All aspects of the ontologies developed by the POC, including the GSO, are a collaborative effort and involve evaluation and assessment by numerous external experts. Before each ontology is released to the public, the POC's internal board of senior editors provides critical assessments and offers suggestions for substantive changes that are thoroughly discussed and incorporated into a revised version of the ontologies. The revised ontologies are then released to database curators and developers, who check for inconsistencies and provide critical feedback about problems and/or advantages associated with use of the new ontologies. In the final phase, the ontologies are subjected to review (http://www.plantontology.org/docs/growth/growth.html) by an external panel of experts. Over 15 outside scientists with expertise in the growth and development of diverse plant species have provided valuable input to the development of this ontology (http://www.plantontology.org/docs/otherdocs/acknowledgment_list.html).
Ontology Editing Tools and Web Interface
The plant ontologies are built and maintained using the DAG editor (DAG-edit) developed by the GO software group. It is open source software implemented in Java and installed locally; flat files are used to store the ontologies. DAG-edit permits creating and deleting new terms, and adding synonyms in categories such as exact, broad, narrow, or related synonyms. This software also supports a user-defined plug in for reading, saving, importing, and exporting (Harris et al., 2004; http://sourceforge.net/project/showfiles.php?group_id=36855). The ontologies are shown using a tree structure. As the GSO is a relatively small ontology, the DAG-edit shows a good overview of the expanded tree in one window. The tool DAG-edit was superseded by the Open Biomedical Ontology Editor (OBO-edit) in its recent release by the GO software group. The same will be used in the future development and maintenance of the GSO.
The PO uses the Amigo ontology browser as the Web interface for searching and displaying the ontologies (Fig. 4). Querying can be done using term names, numerical identifier, synonyms, or definitions. The associated annotations to terms from all the represented databases can be viewed on the term detail page (Jaiswal et al., 2005).