Terry MeehanScientific Curator
Mouse Genome InformaticsThe Jackson Laboratory
Logical Definitions for Hematopoietic Cell Terms
A Brief History of Hematopoietic Cell Terms in the CL
• CL initially developed in 2003 by Jonathan Bard, David States, Michael Ashburner, and Seung Rhee
• 1st round of revisions to hematopoietic cell terms in 2006• adjunct to major improvements to the Gene
Ontology• 80 new cell types• Is_a parentage provided for all
hematopoietically derived cells• Develops_from parentage provided for all major
hematopoietic lineages
A Brief History of Hematopoietic Cell Terms in CL
• 2nd round, 2008 - 2009 – NIAID Cell Ontology Workshop (May 2008).
• Gathering of NIAID experts on immune system cells with an interest in ontology
• Two day meeting focusing on:• Focus on T cells, B cell, dendritic cells, macrophages• Ontology best practices
• Two publications• “An improved ontological representation of dendritic
cells as a paradigm for all cell types” Masci et al., BMC Bioinformatics 2009
• “Hematopoietic cell types: Prototype for a revised cell ontology” Diehl AD et al., J Biomed Inform. 2010
Current Development of Hematopoietic Cell Terms
• 3rd round, 2009 - present – ARRA funded improvements to Cell Ontology as supplement to GO grant
• Building upon outcome of the NIAID Cell Ontology Workshop
• 1st year focus is upon using the hematopoietic cells as a test case for redevelopment of the entire Cell Ontology• Create “Hemo_CL” sub-ontology• Provide logical definitions based on cross-products
using OBO Foundry candidate ontologies
Hemo_CL subontology
Also:- removed insect cell types- DC_CL = dendritic cell ontology- Granulocyte ontology- New terms from literature
Within a couple of weeks, we started generating logical definitions for this sub-ontology
Logical Definitions (cross-product)• Genus-differentia approach• “an X is a G that D”• X = class we are defining • G = genus (is_a parent) • D = differentia, the characteristics that
discriminate instances of X from other instances of G
“CD4+, alpha-beta T cell” is a “alpha-beta T cell” that has_plasma_membrane_part CD4X G D
Ontologies & Relationships used in Differentia
Generating Logical Definitions“Platelet” as it appears in CL
Generating Logical DefinitionsReduce is_a parent count to 1
Generating Logical Definitions
Portion of the OBO stanza
intersection_of: CL:0000763 ! myeloid cellintersection_of: capable_of GO:0007596 ! blood coagulationintersection_of: capable_of GO:0008015 ! blood circulationintersection_of: has_quality PATO:0001405 ! anucleateintersection_of: has_quality PATO:0001874 ! discoid
”Platelet: A non-nucleated disk-shaped cell formed by extrusion from megakaryocytes, found in the blood of all mammals, and mainly involved in blood coagulation.”
Use free text for cross-product formation
Generating Logical DefinitionsInfer relationships with reasoner
Benefits of Using Logical Definitions
Ontology engineer need only place term along one axis of differentia
? ??
Benefits = Connections to Other Ontologies
OntologyRelationshipYellow = PATO D = develops_fromGreen = GO-bp C = capable_ofBlue = Hemo_CL Q = has_quality P = part_of
Red = GO-cellular componentBlue = Hemo_CL
= has_plasma_membrane_part
Benefits = Reasoner flags errors
Declared disjoint_from each other
Benefits = Reasoner flags errors
Corrected logical definition
Benefits = Unexpected associations
Conclusion I
• Hematopoietic cell types have been given logical definitions using terms from other OBO Foundry candidate ontologies
• Logical Definitions allow use of automated reasoners to • infer relationships along other axes of differentia
within CL• infer relationships to other OBO Foundry
candidate ontology• Flag potential errors in logic• Find unexpected associations between terms
Problems Solved- Species differences
• Hematopoietic cell types described by cell surface markers
• Cell types characterized by presence, absence, or differing levels of expression of cell surface markers
• Cell surface markers differ between species
• Solution: Issues encountered in the DC_CL• new relationships (next talk)• creation of species-specific (-biased?) terms
New Developmental Cell Types
22 new developmental cell types added
Problems Solved–Different Viewpoints
Flow cytometry= Diwan A et al. PNAS 2007;104:6794-6799histology = http://www.som.tulane.edu/classware/pathology/Krause/Blood/Blood.html
minimal information needed to substantiate a cell typevs different ways of identifying cell type
Problems Solved–Different Viewpoints
• Soln:• Protein and protein complexes found on cell
surface is main axis of differentia• amendable with flow cytometry
• Some terms (macrophages) anatomical location is important• logical definition using UBERON terms
• Histological characteristics captured by new terms that are GO-cell components x PATO terms• ex- Basophilic cytoplasm• 8 terms given cellular-phenotype namespace (CP:)
Problems Solved–Different Viewpoints
• Potential issues • What ontology to put CP terms?• Expression arrays, RNA-seq used to
identify cell types• Moving into neurons• number of CP terms will greatly increase• differing anatomy nomenclature within the
same species and across species
Problems Not Solved–True path violation with high level termsT cell:“A type of lymphocyte whose defining characteristic is the expression of a T cell receptor complex.”
Thymocyte is_a pro-T cell?Pro-T cells are cells “not fully committed to theT cell lineage”
Currently (not completely satisfactory):T cells are lymphocytes capable_of “T cell mediated immunity”
Problems Not Solved- Ontological Gaps• Terms needed for logical definitions
missing• many cases just need to request term• -20+ new terms requested for the GO
• Not clear what ontology should cover the following:• Invariant T cell receptors• Type I NK T cells and mucosal invariant T cell
• carbohydrate/lipid/protein complexes• alpha-gal Cer/ Cd1D• tetramers- OBI?
Ontological groups are currently discussing these issues
Conclusion II
• Generated species-specific cell type terms• Included minimal information for different
points of view• Be aware of true-path violations with high
level terms• Work with other groups on the “ontological
gaps”
Future Steps
• Logical definitions for high level terms• Add back in insect immune cell types• Incorporate into CL• Neurons!