Ontology Concepts

Session Overview

Introduces the concepts, vocabulary, and structure of biological ontologies as used in Chado and the SNP-Seek platform. This lecture provides the conceptual grounding needed before the Chado schema deep dive that follows — tracing the path from a biological idea through an ontology term to a typed record in the database.

Presented by Jeffrey Detras, PhD — Senior Specialist, Bioinformatics.

Topics Covered

1. Motivation

• The problem: biology has many names for one idea — gene, Gene, locus, coding_gene
• Controlled terms reduce ambiguity, typos, and local naming habits
• Chado links biological records to controlled vocabulary terms instead of relying on free text

2. Controlled Vocabulary vs Ontology

• Controlled vocabulary — a list of accepted terms; useful for consistency and validation
• Ontology — terms + definitions + relationships; useful for meaning, integration, and broader queries
• A controlled vocabulary standardizes words; an ontology standardizes meaning

3. Anatomy of a Term

• Name — readable label (e.g. gene)
• Stable ID — persistent reference (e.g. SO:0000704)
• Definition — limits ambiguity (e.g. sequence region with function)
• Relationships — places the concept in context (e.g. is_a biological_region)
• Synonyms — helps map alternatives (e.g. locus, coding gene)

4. Ontology Structure — Relationships as a Biological Graph

• The power of an ontology comes from edges between concepts
• Common relationship types: is_a (subtype), part_of (component), derives_from (origin), regulates (process)
• Chado stores relation labels as cvterms too, allowing the vocabulary of relationships to be extended
• Reasoning: a query for a broad parent term can include records annotated to more specific child terms

5. Ontologies Commonly Used Around Chado

• SO — Sequence Ontology: feature types such as gene, mRNA, exon, CDS
• GO — Gene Ontology: function, process, cellular component
• RO — Relation Ontology: part_of, derives_from, regulates
• TO / PO — Trait and Plant Ontologies: traits, phenotypes, plant structures
• CO — Crop Ontology: CGIAR community-developed ontologies
• Local CV — project-specific vocabulary for terms not covered by public ontologies
• Principle: public ontology when possible; local CV when needed

6. The Chado CV Module

• cv — the ontology or vocabulary namespace
• cvterm — the term or concept; the central object that many other tables point to
• dbxref — the external identifier (e.g. SO:0000704)
• cvtermsynonym — alternative names for a term
• cvterm_relationship / cvtermpath — direct and transitive relationships between terms

7. cvterm as the Semantic Bridge

• Concept pathway: biological idea → ontology term → cvterm → typed Chado record
• Typing a feature as a gene is not just a label — it is a link to a defined ontology concept via feature.type_id → cvterm.cvterm_id
• Example: GFF strings (gene, mRNA, exon) must resolve to matching cvterms on load

8. Practical Implications for Loading and Querying

• Data integrity — avoids misspellings, duplicates, and informal labels
• Portability — shared ontology terms allow datasets to be integrated
• Reasoning — parent-child relationships enable queries across broad concepts
• Troubleshooting — trace loader errors to missing, wrong, or ambiguous terms
• Most cvterm-related failures are term-resolution problems

9. Applied Concept Check — When a Term Does Not Resolve

• Example: loader reports "Could not find cvterm 'lncRNA'"
• Step 1 — What kind of concept is it? (feature type, relationship, or property)
• Step 2 — Which ontology should contain it? (sequence features → Sequence Ontology)
• Step 3 — Is the accepted name different? (lnc_RNA vs lncRNA)
• Step 4 — What to fix? Correct the input, load/update the ontology, or add a local term

📄 Open Slides (PDF)