Revision as of 13:36, 29 May 2015

There are 3 ways of creating “ontologies” on the Crop Ontology website.

• First, one can choose to upload a Trait Dictionary in Excel. The TD is only used to store and load information about traits, thus it cannot be used for uploading ontologies that would belong to the other categories (Note, there are 5 sections on the CO website: General Germplasm Ontology, Phenotypes and Traits Ontology, Structural and Functional Genomic Ontology, Location and Environmental Ontology and Plant Anatomy & Development Ontology).
• OBO files can be uploaded as well. They can be used to create ontologies in any of the 5 sections.
• Finally, ontologies can be built from scratch using the website dedicated interface.

Once the files are uploaded or some terms are submitted using the web interface, the data are stored in a noSQL database (Google App Engine Datastore). The ontologies can be downloaded in various formats: CSV, OBO, SKOS and JSON. The structures of the different formats are described in the following sections. Note that because there is no exact mapping between the different formats some compromises has been made. Not all the information is available in the different formats.

Trait Template

Structure of the Trait Template V4

So far, the Trait Template V4 is uploaded into the CO website. The Trait Template is used to gather information on traits along with their methods and scales of measurement. For each trait, some meta-information is added (5 first columns). An (unique) ID is created by the system for each trait, method and scale (several IDs are created when the scale is categorical). But for now, if two methods or scales are the same (seed length and leaf length are measured in centimeters), different IDs are created, which is far from being ideal or even ontologically correct. The Trait Template V4 columns are:

• IBFieldbook
• Name of submitting scientist
• Institution
• Language of submission (only in ISO 2 letter codes)
• Date of submission
• Crop
• Name of Trait
• Abbreviated name
• Synonyms (separate by commas)
• Trait ID for modification, Blank for New
• Description of Trait
• How is this trait routinely used?
• Trait Class
• Method ID for modification, Blank for new
• Name of Method
• Describe how measured (method)
• Growth Stage
• Bibliographic Reference
• Comments
• Scale ID for modification, Blank for new
• Type of Measure (Continuous, Discrete or Categorical)
• For Continuous: units of measurement
• For Continuous: reporting units (if different from measurement)
• For Continuous: minimum
• For Continuous: maximum
• For Discrete: Name of scale or units of measurement
• For Categorical: Name of rating scale
• For Categorical: Class 1 - value = meaning
• For Categorical: Class 2 - value = meaning
• For Categorical: Class 3 - value = meaning
• For Categorical: Class 4 - value = meaning
• For Categorical: Class 5 - value = meaning

OBO files

OBO file structure

OBO files are created using the obo2owl library (which should sound familiar ;-)). The main issue with the Trait Template is that because different IDs can be created for the same “thing” (e.g. the term “cm” will get different IDs if different traits are measured in cm), different obo terms can have the same name (but different IDs). That is a problem when our users tried to import the OBO files in a Chado database. Consequently, I created a little script that created OBO files where all the names are unique in a given namespace. The script has not be inserted in the website yet but it will be soon. Example of an OBO files on Github.

SKOS files

SKOS file structure

List of Crops and IDs

According to the project timeline, the crops that will be linked to the reference ontologies are:

The trait ontologies start with 3XX whereas the anatomy ontologies start with 1XX.