Part 1: Creating your first LinkML schema

We assume that you already have LinkML installed

For the purposes of this tutorial, the simplest setup is to use a virtual environment, and then install linkml:

mkdir linkml-tutorial
cd linkml-tutorial
python3 -m venv venv
source venv/bin/activate
pip install linkml

You can check the install worked:

linkml-convert --help

As always, you can consult the FAQ if you have issues.

Your first schema

Our first schema consists of a single class Person, with a number of slots:

First create a file

personinfo.yaml:

id: https://w3id.org/linkml/examples/personinfo
name: personinfo
prefixes:
  linkml: https://w3id.org/linkml/
  personinfo: https://w3id.org/linkml/examples/personinfo
imports:
  - linkml:types
default_range: string
default_prefix: personinfo

classes:
  Person:
    attributes:
      id:
      full_name:
      aliases:
      phone:
      age:

(note that all files are available in the examples/tutorial folder of this repository)

Converting to JSON-Schema

Now run the following command on the file you just created:

gen-json-schema personinfo.yaml

Outputs:

{
    "$defs": {
        "Person": {
            "additionalProperties": false,
            "description": "",
            "properties": {
                "age": {
                    "type": "string"
                },
                "aliases": {
                    "type": "string"
                },
                "full_name": {
                    "type": "string"
                },
                "id": {
                    "type": "string"
                },
                "phone": {
                    "type": "string"
                }
            },
            "title": "Person",
            "type": "object"
        }
    },
    "$id": "https://w3id.org/linkml/examples/personinfo",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "additionalProperties": true,
    "metamodel_version": "1.7.0",
    "title": "personinfo",
    "type": "object",
    "version": null
}

Don’t worry if you don’t know much about JSON-Schema. This is just an illustration that LinkML can be used in combination with a number of frameworks.

Creating and validating data

Let’s create an example data file. The file will contain an instance of the class we defined in our personinfo.yaml schema:

data.yaml:

id: ORCID:1234
full_name: Clark Kent
age: "32"
phone: 555-555-5555

Note the values of all these fields are strings, even age; we will later return to this example and show how we could model this more naturally as a number.

Validate:

linkml-validate -s personinfo.yaml data.yaml

You should see a message indicating there were no validation issues found. Success!

To see an example of data not validating:

bad-data.yaml:

id: ORCID:1234
full_name: Clark Kent
age: "32"
phone: 555-555-5555
made_up_field: hello

linkml-validate -s personinfo.yaml bad-data.yaml

This should report an error to the effect that made_up_field is not allowed.

Working with JSON

One of the advantages of LinkML is the same datamodel can be used for multiple expressions of the same data, for example:

• YAML/JSON

• TSVs/spreadsheets

• RDF

• Relational Databases

There are various complexities involved in going between these, but YAML and JSON are basically interchangeable with LinkML

data.json:

{
 "id": "ORCID:1234",
 "full_name": "Clark Kent",
 "age": "32",
 "phone": "555-555-5555"
}

This will validate in the same way as the equivalent YAML file:

linkml-validate -s personinfo.yaml data.json

Converting to RDF

You can use the linkml convert tool to convert your data to other formats, including RDF:

linkml-convert -s personinfo.yaml data.yaml -t ttl

This will produce RDF/turtle output as follows

@prefix personinfo: <https://w3id.org/linkml/examples/personinfo> .

[] a personinfo:Person ;
    personinfo:age "32" ;
    personinfo:full_name "Clark Kent" ;
    personinfo:id "ORCID:1234" ;
    personinfo:phone "555-555-5555" .

To write the RDF/turtle output directly to a file use

linkml-convert -s personinfo.yaml data.yaml -o data.ttl

If you are not familiar with RDF that’s OK! RDF is just one of the possible ways of working with LinkML.

If you are familiar with RDF, the first thing you will likely notice is that we are not reusing standard URIs for our properties and classes. Don’t worry! We will get to this later.

Exercises

1. Extend the example schema to include fields for occupation and employed_by

2. Create a test data instance to indicate Clark Kent has an occupation of reporter and is employed by the Daily Planet

3. Validate the data

Next: Collections of data

Our toy example so far has a single person instance. Next we’ll see how to exchange lists of records.