COB: A Core Ontology for Biology

COB: A Core Ontology for Biology Ontology Summit 2023 https://ontologforum.org/index.php/OntologySummit2023 Chris Mungall Lawrence Berkeley National Laboratory @chrismungall https://genomic.social/@cmungall 10.5281/zenodo.7654926 We have many identifiable things and categories Drugs 10k Chemicals 4tn? Species ~9 million Diseases and Phenotypes 10-50k/species Cells 10,000s+ types per species) Experiments Raw data ?? exabytes Genes 20k per species Metagenome dbs: >65bn genes Genetic variants 3m in human alone The things are interconnected across scales Different perspectives on the same things static anatomy static cell structures static disease state static chem structures Different perspectives on the same things dynamic physiology dynamic pathways dynamic pathophysiology dynamic reactions ↑bilirubin → jaundice Conjugation of bilirubin occurs-in liver OBO is for organizing the things http://obofoundry.org 1. Well-integrated Modular ontologies 2. Technical and sociotechnological framework for cooperation 4. Allow us to categorize, organize, and integrate all of the things 3. Tools, best practices and infrastructure for forging new ontologies @obofoundry OBO registry https://obofoundry.org OBO ontologies interoperate 15k citations/year 1 billion annotations 10-1000 billion annotations Millions of environmental samples annotated Diagnosing rare disease patients uses Analyzing single-cell seq You’re welcome! Layers of OBO interoperability Common formats and tools ROBOT, OWL API, rdflib, fastobo, owltools, dosdp-tools etc are able to read and write ontologies in commonly understood formats. 3 FA[IR]ness & Openness Ontologies should be findable, accessible and openly available. 4 Shared design patterns DOSDP templates, ROBOT templates, OTTr and other systems serve as tools to abstract modelling patterns and share them across ontologies. 1 Shared vocabularies and upper level integration RO standardises the relationships to be used in OBO ontologies. OMO standardises the annotation properties to be used for term and ontology metadata. COB provides the upper layer for biological and biomedical ontologies. Term-reuse across OBO ontologies. 2 Credit: Nico Matentzoglu Axiomatizing relations using classes Original RO RO development expressed in ?? ?? occurs in ?? ?? Axiomatizing relations using classes Original RO RO development expressed in material anatomical entity gene occurs in Independent continuant process Root nodes of domain OBOs Original RO RO development expressed in material anatomical entity gene occurs in Independent continuant process SO CARO Abstract relations and classes Original RO RO development BFO expressed in material anatomical entity gene occurs in Independent continuant process SO BFOBFO CARO Basic Formal Ontology Original RO RO development BFO ●Shared philosophical abstractions ● Not intended as a biological top layer Multiple abstract aspects of concept Original RO RO development BFO https://douroucouli.wordpress.com/2022/08/10/shadow-concepts-considered-harmful/ Just tell me where to place my anatomy terms The lack of a consensus biological top level has led to duplicative efforts across ontologies in different branches Problem: leaky abstractions OBI term Upper ontology terms I just want to explore assay terms Problem: level of abstraction is too high I just want to explore assay terms Desiderata for a biological upper core •should provide a parent to every OBO ontology class •should be anchored in BFO, but hide its complexity from end-users •should include logical axioms that make inconsistencies within and between ontologies apparent through reasoning Original RO RO development BFO BFO COB: Common Ontology for Biology Original RO RO development BFO COB COB development process • Workshops ○ 2018 RO meeting (Denver) ○ 2019 ICBO (IRL) ○ 2020 ICBO (Virtual) ○ 2021 ICBO (Virtual) • Slack channel • Modern GitHub-based workflows ○ See ODK talk Feb 8 • Goals: ○ No behind-doors decision making ○ All decisions transparent on issue tracker ○ Anyone can make a PR Practicality, use cases >> philosophy, perfectionism Open: community can make PRs role entity continuant occurrent process biological process planned process assay independent continuant characteristic generically dependent continuant specifically dependent continuant material entity immaterial entity information document color anatomical space organism molecule cell gross anatomical structure cellular process realizable entity 34 COB and BFO Flattened view of BFO for COB •Include only terms that have OBO subclasses •Use bfo terms that are non-threatening for biologist ontology consumers •Keep full logical axiomatization ‘in the background’ process material entity site information characteris tic realizable entity COB – Material Entities Generic: material entity Physical world: subatomic particle, atom or ion, molecular entity, macromolecular entity Biological: cellular component, cell, gross anatomical structure, organism, population Human activities: processed material (not shown) subatomic particle electron atom or ion molecular entity macromolecular entity cellular component cell gross anatomical structure organism population nickel aspirin actin cell nucleus macrophage lung human biocurators smaller larger examples 36 Generic: process, quality, realizable Physical world: site, geographic location, mass, color Biological: biological process, anatomical space, role Human activities: planned process, information COB – Non-Material Entities process site quality characteristic information biological process planned process anatomical space geographic location mass color role plan document measurement datum 37 COB – Examples cellular component organelle cell nucleusmaterial entity subatomic particle nuclear particle atomic nucleus gross anatomical structure organ nucleus of brain material entity organism processed material cloning vector infectious disease vector "nucleus" "vector" process planned process data transformation support vector machine 38 In progress: phenotype and disease • Upper ontology assumptions don’t align with actual uses of phenotype and disease ontologies • Consider Schulz conflation model? In progress: units and measurements Proposed simplification of existing measurement model Biolink-Model: A schema for biological KGs ● Expressed in LinkML ● “Ontology-like” ● OBO classes are instances ● Edges are first-class Alignment with Biolink https://biolink.github.io/biolink-model Developed as part of NCATS Translator project ● Weekly Data Modeling calls (20-40 people) ● Working groups for different areas (e.g. chemicals) ● Technically diverse group (domain scientists, bioinformaticians, ontologists) ● Use of GitHub (PRs, votes)