Skip to content

Latest commit

 

History

History
106 lines (89 loc) · 5.94 KB

WORKFLOW_NEW.md

File metadata and controls

106 lines (89 loc) · 5.94 KB
Raw

Creating a New Ontology

These instructions provide the basic steps for creating a new trait ontology. The ontology will be managed in an Excel Workbook (referred to as a Trait Workbook) which will be used to generate a Trait Dictionary (used to update the croptontology.org website) and a SGN-OBO File (used to load the ontology into a breedbase instance).

Create a New Trait Workbook

You can create a mostly blank Trait Workbook using the create_tw.pl script. You'll need to give the script the following parameters:

  • -d = The default ontology namespace. This namespace will be applied to all of the ontology terms loaded into breedbase. (tomato_trait)
  • -n = The human-readable name for the ontology. This name will be displayed on the bredebase website. (Tomato Traits)
  • -r = The root ontology id, such as the Crop Ontology Identifier. If you're creating a new ontology and this does not yet exist, you'll first need to get an ID from the curators of the cropontology.org website. If you don't have one yet, you can use a placeholder (such as CO_999) and update it in the Trait Workbook later. (CO_999)
  • -o = The path to the generated Trait Workbook .xlsx file. (./traits.xlsx)

When not providing the input argument, the script will create a new trait workbook file containing a single variable, trait, method and scale as an example.

perl create_tw.pl -d tomato_trait -n "Tomato Traits" -r CO_999 -o ./traits.xlsx -v

Ontology Term Definitions

At this point, you should have a mostly blank Trait Workbook. Each worksheet in the workbook contains information for each of the different types of ontology terms. It is important to know how the different ontology terms are defined and how they relate to each other:

  • A variable is the term that actual phenotype observations are associated with. The variable is composed of a trait, method, and scale. For example, when you go out into the field and measure plant height you will associate your measurement with a single variable that is composed of the trait for plant height, the method that describes the way you observed the trait (such as a direct measurement of plant height using a meter stick), and the scale used to record the measurmenet (such as cm).
  • A trait is the entity that is observed. In our plant height example, the trait would be plant height with a description of how plant height is defined for your crop.
  • A method is the procedure for observing a trait. In our example for plant height we would create a method for the direct measurement of plant height that describes how to measure plant height directly in the filed using a meter stick. A trait can have more than one method associated with it, if it is observed in different ways (such as the direct measurement of the height using a meter stick vs a calculated measurement of plant height using a clinometer).
  • A scale is the units in which a trait is observed (such as cm). A categorical scale can have descriptions for each of its categories.
  • A trait class is a category of traits. Each trait should be associated with a single trait class to keep similar traits together (such as Agronomic, Quality, Disease, etc).
  • The single root term contains the Crop Ontology ID, the ontology name and its namespace. This information will be used when generating the trait dictionary and obo files.

Trait Workbook Usage

Each ontology data type has a separate worksheet in the Trait Workbook. Almost all of the columns come directly from the Crop Ontology Trait Dictionary format. For more information and column descriptions, look at the README sheet in the Trait Dictionary template (such as the Trait Dictionary template v5) found on the cropontology.org website.

There are some columns that are not part of the Trait Dictionary format. These are found on the Variables worksheet:

  • Variable label: this is used as the name of the variable term in the generated obo files. This name is then displayed as the variable name on the breedbase website. In the Trait Dictionary of existing crops, the Variable name column often uses an abbreviated / coded form of the variable name (such as PH_M_cm) and is not very user-friendly for users of the breedbase website. The Variable label column should use a human-readable name that describes the variable in a concise yet descriptive way.
    We have found that the form {trait name} - {scale name} (such as 'Plant Height - cm') works well for most variables. If you have a trait that uses the same scale but different methods, you will need to include some short description of the method in the label name to differentiate them (such as 'Plant Height - measured - cm' and 'Plant Height - calculated - cm').
  • The Trait name, Method name, and Scale name columns are used to identify the trait, method, and scale that compose the variable. All three terms must be defined and must match the name of the term exactly as defined in their corresponding worksheets.
  • The VARIABLE KEY column concatenates the trait, method, and scale names to check for duplicates of the same combination of trait, method and scale. There should only be one variable for each combination of trait, method and scale.

IDs

  • The Variable ID, Trait ID, Method ID, and Scale ID columns need to have an ID number that is unique across all of the data types (a scale and a method cannot have the same ID number). This column should contain an integer and will be expanded into the full term ID (such as CO_999:0000001) when building the trait dictionary and obo files.

Building Files

Once you have a Trait Workbook with your trait information filled in, you can follow the steps outlined in the Loading an Existing Ontology into Breedbase workflow to create an obo file and/or trait dictionary and load the obo file into a breedbase instance.