Skip to content

MiCoKnowGraph

A draft data model for a microbiome knowledge graph (KG), written as a starting point for BioHackathon Europe 2026, Project 16. Source: https://github.com/FatemehFN/micokg-blueprint.

Where to start

You are… Read
New to the project Onboarding: set up and run everything
New to RDF, ontologies or SPARQL Knowledge-graph primer
Adding classes, slots or data Modelling guide
A domain expert with questions the KG should answer Use cases
Working on mappings or VERDANT VERDANT integration
Wondering why something is modelled the way it is The design notes, starting with associations as nodes

The design in one picture

flowchart LR
  subgraph Entities["Entities (ontology IRIs)"]
    T[Taxon<br/>NCBITaxon]
    D[Disease<br/>MONDO]
    M[Metabolite<br/>ChEBI]
    G[Gene / AMR determinant<br/>GO, ARO, SO]
    F[Food<br/>FoodOn]
  end
  S[Sample<br/>MIxS + ENVO/UBERON] -->|hasFeatureOfInterest ◀| O[Observation<br/>SOSA]
  O -->|is about| T
  O -->|is about| G
  A((Association<br/>direction, site,<br/>evidence, study)) -->|rdf:subject| T
  A -->|rdf:object| D
  A -->|derived from| O
  X[[SSSOM mappings]] -.->|exactMatch| D
  • Observations are raw per-sample measurements.
  • Associations are claims made from them (or taken from the literature) and carry their context and evidence.
  • Mappings connect identifiers from different ontologies, so a DOID-coded record from one resource still answers a MONDO-coded question.