Modularity requirements in bio-ontologies: a case study of ChEBI

EBI is an Outstation of the European Molecular Biology Laboratory.

Modularity requirements in bio-ontologiesa case study of ChEBI

Janna Hastings, Colin Batchelor,

Stefan Schulz, Christoph Steinbeck

Workshop on Modular Ontologies, ESSLLI,12 August 2011

ChEBI: an ontology of biologically interesting chemicals

12.04.20232

ChEBI Ontology

chemical entity role

carboxylic acid

application

antibacterial drug

cefpodoxime (CHEBI:606443)

pharmaceutical

biological role

chemical role

chemical substance

molecular entitygroup

carbonyl compoundsolvent

has role

cyclooxygenaseinhibitor

carboxy group

has part

Bio-ontologies are modular by design:domain and granularity

ChEBI ontology12.04.20233

Chemistry

Molecular entities

Substances

Functions and roles of

chemical entities

Domain

Granularity Upper level type

Material entities

They are characterised by large sizes and low expressivity


Chemical entities (29132)Roles (596)Subatomic par-ticles (41)

August 2011 29769 classes in total

Currentlyexportedin EL++

Classification practices in chemistry lead to high levels of multiple inheritance



ChEBI is growing

bigger …

… and more expressiveImage credit: Jonathan J. Dickau

Increased expressivity to enable automatic classification

hydrocarbon equivalentTo

molecule and has_atom only (carbon atom or hydrogen atom)

peptide cation equivalentTo

peptide and has_charge some double [>, 0.0]


Value restriction

Data range restriction


carboxylic acid equivalentTo

molecule and has_functional_group some carboxy group

tricarboxylic acid equivalentTo

molecule and has_functional_group exactly 3 carboxy group

Cardinality restriction

Size explosion in asserted parts


Reasoning is required for classification and consistency validation

No definitional cycles

A part_of B part_of C part_of A

Enforcing disjointness

Chemical Entity disjoint_from Role …

Group disjoint_from Molecule …

No disallowed combinations of relations

A has_part B ; A conjugate_base_of B



Number of fully defined classes

Rea

soni

ng t

ime

in s

econ

dsHermiT classification of ChEBI


Modularity and large ontologiessmaller modules = faster classification

smaller modules = easier maintenance


A USEFUL module for maintenance

… is delineated by topic

… is comprehensible and easy to work with

… is self contained for reasoning tasks

Subject-specific modules overlap


Drugs

Metabolism

Immunology

Carboxylic acids

E.g. GO-SLIM approach

Self-contained modules include all axioms needed for

classification and consistency checking


parts

propertiesupper-level constraints(e.g. disjointness)

hierarchy

Ontology segmentation tools don’t work very well on ChEBI

… yet


Modules too small or too big

Out of memoryLong processing times

No tool supportfor recombined

viewing/querying

Topic blind

ChEBI ontology12.04.202317Image credit: Rameesh Vyas

Interrelating bio-ontologies requires

modular imports

The MIREOT mechanism requires manual selection of module content

and manual update of ontology changes


Choose terms

Extract module

Build ontology

links

We need modular ontology views


Ontology O

View V1 (Topic, Editing)

Automatic module extractionbased on selection criteria

Edit, Validate,write back to source

Views can be imported and arethen automatically updated


Ontology O1 (e.g. chemistry)

View V1 (Topic, Editing)

Module extraction

Ontology O2 (e.g. biology)

Import of views

Automaticupdate


How do we facilitate

the development of tools

for modular ontology engineering?

EBI is an Outstation of the European Molecular Biology Laboratory.

Acknowledgements: BBSRC (funding)

Thank you

12.04.2023

ChEBI ontology22

Modularity requirements in bio-ontologies: a case study of ChEBI

Technology

role cefpodoxime chebi

chebi ontologyimage

chebi yettopic blind

role group

modular ontologies

useful module

selfcontained modules

chemistry lead