Ewance
← All challenges AI Solutions Architect

Build an OWL Ontology for a Pharma R&D Knowledge Base

Duration
3 weeks
Level
Advanced
Type
Design
Team
1 to 5 people

The situation

The pharma-AI startup (around 50 people, working with mid-cap pharma clients) is replacing its first-generation graph database with an ontology-backed system and needs the new substrate to satisfy both the chemistry and the medical-affairs teams without endless review cycles.

Your task

Design and populate an OWL ontology for compound-target-assay relationships that supports reasoner-derived inferences and an ongoing stewardship process.

The brief

You start with

You receive a CSV-form starter knowledge base (around 4,000 compounds, 600 targets, 1,200 assays) and a list of 12 competency questions the scientists currently can't answer with their database.

What you do

  1. Design an OWL ontology in Protege covering at minimum: Compound, Target, Assay, Mechanism, Indication, plus 8-12 named properties.
  2. Reuse ChEBI/Mondo/Uberon classes where appropriate (link via owl:equivalentClass or rdfs:subClassOf).
  3. Populate the ontology with the starter data, run a reasoner (HermiT or ELK), and demonstrate 3 useful inferred facts.
  4. Write the stewardship playbook covering versioning, change review, and external-ontology resync.

What you will practice

  • Design an OWL ontology aligned to real competency questions
  • Use description-logic reasoners to derive inferred facts
  • Integrate external ontologies (ChEBI, Mondo, Uberon) without duplication
  • Document an ontology stewardship process for ongoing maintenance

A suggested way in

  1. Start from competency questions; classes and properties should fall out of what scientists need to ask
  2. ELK reasoner is much faster than HermiT for the OWL 2 EL subset
  3. Version the ontology like code: every change is a PR with explanation

Skills you will practice

Ontology design, Owl, Knowledge representation, Description logics, Reasoning, Protege

Where it leads

AI Solutions Architect, AI Engineer, Data Engineer

How it is assessed

  • Ontology coverage30%

    Competency-question coverage is high and the model is parsimonious (no orphan classes).

  • Reasoner inferences25%

    Three inferred facts are useful and explained via the reasoner's justification trace.

  • External alignment20%

    ChEBI/Mondo/Uberon links are correct and non-duplicating.

  • Stewardship playbook25%

    Playbook covers versioning, change review, and resync of upstream ontologies.

Resources

  • ProtegeTool

    Standard OWL ontology editor with reasoner integration.

  • ELKTool

    Fast OWL 2 EL reasoner for the populated ontology.

  • HermiTTool

    Backup reasoner for OWL 2 DL features ELK doesn't cover.

  • PythonTool

    Use Owlready2 or rdflib to load the starter CSV into RDF triples.

  • GitTool

    Version the ontology with proper change history.

  • NotionTool

    Author the stewardship playbook and competency-question coverage table.

Sample Ewance certificate of completion, verified by LearnCoin
Sample certificate
Your proof

Finish it. Earn this.

A verified certificate with your name. Anyone can check it.

Time to practice.

Solve this challenge
Related challenges

More like this.