Introduction to Graph Data and Ontologies

Question 1

What is Data integration?

Answer 1

The analysis of genomic, imaging or other
types of data allows us to investigate different
facets of human health.
• But in order to gain a comprehensive
understanding of human health, we need to
integrate such data

Question 2

What are the challenges to data integration?

Answer 2

1. Biomedical and healthcare datasets sit in
   silos.
2. Linking entities between different datasets is
   not a trivial task.
    In Scotland, we use CHI Numbers to uniquely
   identify patients.
    But how about sharing data between different
   countries?
3. Ambiguity around the meaning of
   different terms.

Question 3

What are graph databases?

Answer 3

Graph Databases use graph structures with
nodes, edges and properties to represent and
store data

Question 4

What are nodes?

Answer 4

T

Question 5

What are edges?

Answer 5

T

Question 6

What are properties?

Answer 6

T

Question 7

What is the RDF graph data model?

Answer 7

Data is represented in the form of triples, i.e.
statements consisting of a subject, a predicate
and an object.
-Subject
-Predicate
-Object

Question 8

Describe RDF triple visualisation

Answer 8

T

Question 9

What are URIs?

Answer 9

In RDF, we use URIs (Uniform Resource
Identifiers) to uniquely identify concepts and
entities.
• Examples:
 http://dbpedia.org/resource/Edinburgh
 http://xmlns.com/foaf/0.1/age
• URIs are used for both resources and
properties.

Question 10

How to use existing URIs

Answer 10

DBPedia (http://dbpedia.org) is a very good source of URIs.
• Every resource that is the subject of a page in Wikipedia has a
corresponding URI in DBpedia.
• URI forEdinburgh:
http://dbpedia.org/resource/Edinburgh

Question 11

How to create your own URIs

Answer 11

If you don’t own a domain name, you can use
http://example.com/
http://example.com/id/EwanMcGregor
 Keep it simple

Question 12

How to merge RDF data

Answer 12

By uniquely identifying resources with the use
of URIs, we can easily link data about the
same resource.
• Merging different RDF datasets is simply a
matter of bringing the two sets of RDF
statements together
Dataset3 = Dataset1 + Dataset2

Question 13

How to write RDF statements in Turtle

Answer 13

Turtle (Terse RDF Triple Language): One of the
most popular forms of syntax for expressing
RDF.
• General form:
subject predicate object

Question 14

What is Turtle?

Answer 14

Turtle (Terse RDF Triple Language): One of the
most popular forms of syntax for expressing
RDF
Whitespace and full stop
When using URIs, these should be enclosed in
angle brackets, e.g.

Question 15

What is ontology?

Answer 15

A formal, explicit
specification of a shared conceptualisation.
• Essentially, a way of encoding domain
knowledge.
• Something like an enhanced dictionary, where
you can look up the meaning of different
concepts and find relations between them.

Question 16

What are the components of ontology?

Answer 16

Classes (e.g. Woman)
 Individuals (e.g. Lucy)
 Attributes (e.g. Age)
 Relations (e.g. MotherOf)
• Ontologies often contain a class taxonomy.
• Formal definitions of classes may also be
included.

Question 17

Why are ontologies useful?

Answer 17

• Allow us to attach meanings to data
• Enable standardisation of terminology
• Allow us to infer new knowledge from existing data

Question 18

What is Gene Ontology?

Answer 18

 It represents information about biological
processes, cellular components and molecular
functions.

Question 19

What is Disease Ontology?

Answer 19

 It provides descriptions of human disease terms,
phenotype characteristics and related
medical vocabulary disease concepts

Question 20

What is SNOMED-CT?

Answer 20

It is a collection of medical terms. It includes
codes, terms, synonyms and definitions used in
clinical documentation and reporting.
 It is considered to be the most comprehensive,
multilingual clinical healthcare terminology in the
world.