10:30-12:00 How to Build an Ontology 1-2pm Best Practices and Lessons Learned 2-3pm BIRN Ontologies: An Overview

10:30-12:00 How to Build an Ontology 1-2pm Best Practices and Lessons Learned 2-3pm BIRN Ontologies: An Overview

http://ontology.buffalo.edu/smith2

How to Build an Ontology


High quality shared ontologies build communities

General trend on the part of NIH, FDA and other bodies to consolidate ontology-based standards for the communication and processing of biomedical data.

NCIT / caBIG / NECTAR / BIRN / OBO ...


TWO STRATEGIES:

Ad hoc creation of new database schemas for each research group / research hypothesisvs.

Pre-established interoperable stable reference ontologies in terms of which all database schemas need to be defined


How to create the conditions for a step-by-step evolution towards gold standard reference ontologies in the biomedical domain

... and why we need to create these conditions

OBO Core project


Ontology =def

A representation of the types of entities existing in a given domain of reality, and of the relations between these types


Types have instances

Ontologies are like science texts: they are about types

(Diaries, databases, clinical records are about instances)


The need

strong general-purpose classification hierarchies created by domain specialists clear, rigorous definitionsthoroughly tested in real casesontologies teach us about the instances in reality by supporting cross-disciplinary (cross-ontology) reasoning about types


The actuality (too often)

myriad special purpose ‘light’ ontologies, prepared by ontology engineers and deposited in internet ‘repositories’ or ‘registries’


these light ontologies often do not generalize …

repeat work already done by others

are not interoperable

reproduce the very problems of communication which ontology was designed to solve

contain incoherent definitions

and incoherent documentation


BIRN Ontology Experiences

In the short-term, users will probably download the data or analyses and extract the results using their preferred methods.

In the long term, however, that will become infeasible– the databases will have to be made interoperable with

standard datamining software.

This is where the neuroanatomy ontologies come in. – We will need to know what the ROI is and which naming

scheme it came from (e.g., a Brodmann’s area, or a sulcal/gyral area, etc.). We’ll need to know how it was defined (Talairach atlas? MNI atlas? LONI atlas? Or subject-specific regions?) and what the statistic is.


BIRN Ontology Experiences

In the short-term, users will probably download the data or analyses and extract the results using their preferred methods.

In the long term that will become infeasible


The long term begins here


A methodology for quality-assurance of ontologies

tested thus far in the biomedical domain on:

– FMA– GO + other OBO Ontologies– FuGO– SNOMED– UMLS Semantic Network– NCI Thesaurus– ICF (International Classification of Functioning,

Disability and Health)– ISO Terminology Standards– HL7-RIM



accepted need for application of this methodology:

– FMA– GO + other OBO Ontologies– FuGO– SNOMED– UMLS Semantic Network– NCI Thesaurus– ICF (International Classification of Functioning,

Disability and Health)– ISO Terminology Standards– HL7-RIM



signs of hope:

– FMA– GO + other OBO Ontologies– FuGO– SNOMED– UMLS Semantic Network– NCI Thesaurus– ICF (International Classification of

Functioning, Disability and Health)– ISO Terminology Standards– HL7-RIM


We know that high-quality ontologies built according to this methodology can help in creating high-quality mappings between

human and model organism phenotypes


“Alignment of Multiple Ontologies of Anatomy: Deriving Indirect Mappings from Direct Mappings to a Reference Ontology”

Songmao ZhangOlivier Bodenreider

AMIA 2005


We also know that OWL is not enough to ensure high-quality ontologies

and that the use of a common syntax and logical machinery and the careful separating out of ontologies into namespaces does not solve the problem of ontology integration


A basic distinction

type vs. instance

science text vs. clinical document

man vs. Musen


Instances are not represented in an ontology

It is the generalizations that are important

(but instances must still be taken into account)


A 515287 DC3300 Dust Collector Fan

B 521683 Gilmer Belt

C 521682 Motor Drive Belt


Ontology Types Instances


Ontology = A Representation of Types



Each node of an ontology consists of:

• preferred term (aka term)

• term identifier (TUI, aka CUI)

• synonyms

• definition, glosses, comments



Nodes in an ontology are connected by relations:

primarily: is_a (= is subtype of) and part_of

designed to support search, reasoning and annotation


siamese

mammal

cat

organism

substancetypes

animal

instances

frogleaf class


Rules for formating terms• Terms should be in the singular• Terms should be lower case• Avoid abbreviations even when it is clear in

context what they mean (‘breast’ for ‘breast tumor’)

• Avoid acronyms• Avoid mass terms (‘tissue’, ‘brain mapping’,

‘clinical research’ ...)• Each term ‘A’ in an ontology is shorthand for a

term of the form ‘the type A’


Motivation: to capture reality

Inferences and decisions we make are based upon what we know of reality.

An ontology is a computable representation of the underlying biological reality.

Designed to enable a computer to reason over the data we derive from this reality in (some of) the ways that we do.


Concepts

Biomedical ontology integration will never be achieved through integration of meanings or concepts

The problem is precisely that different user communities use different concepts

Concepts are in your head and will change as your understanding changes


Concepts

Ontologies represent types: not concepts, meanings, ideas ...

Types exist, with their instances, in objective reality

– including types of image, of imaging process, of brain region, of clinical procedure, etc.


Rules on types

Don’t confuse types with wordsDon’t confuse types with conceptsDon’t confuse types with ways of getting to

know typesDon’t confuse types with ways of talking

about typesDon’t confuses types with data about types


Some other simple rules for high quality ontologies


Univocity

Terms should have the same meanings on every occasion of use.

They should refer to the same kinds of entities in reality

Basic ontological relations such as is_a and part_of should be used in the same way by all ontologies


PositivityComplements of types are not themselves types. Hence terms such as

non-mammal non-membrane other metalworker in New Zealand

do not designate types in reality

There are also no conjunctive and disjunctive types:

protoplasmic astrocyte and Schwann cellPurkinje neuron or dendritic shaft


Objectivity

Which types exist is not a function of our knowledge.

Terms such as ‘unknown’ or ‘unclassified’ or ‘unlocalized’ do not designate types in reality.


Single Inheritance

No kind in a classificatory hierarchy should have more than one is_a parent on the immediate higher level


Multiple Inheritance

thing

carblue thing

blue car

is_a1 is_a2


is_a Overloading

serves as obstacle to integration with neighboring ontologies

The success of ontology alignment demands that ontological relations (is_a, part_of, ...) have the same meanings in the different ontologies to be aligned.

See “Relations in Biomedical Ontologies”, Genome Biology May 2005.

DISEASE MAPS


General Rule

Formulate universal statements first

Move to A may be B in such and such a context later


Intelligibility of Definitions

The terms used in a definition should be simpler (more intelligible) than the term to be defined; otherwise the definition provides no assistance

– to human understanding

– to machine processing


Definitions should be intelligible to both machines and humans

Machines can cope with the full formal representation

Humans need clarity and modularity


But

Some terms are primitive (cannot be defined)AVOID CIRCULAR DEFINITIONS

Avoid definitions of the forms:

An A is an A which is B (person = person with identity documents)

An A is the B of an A (heptolysis = the causes of heptolysis)


Case Study: The National Cancer Institute Thesaurus (NCIT)

does not (yet) satisfy these and other simple principles


The NCIT reflects a recognition of the need

for high quality shared ontologies and terminologies the use of which by clinical researchers in large communities can ensure re-usability of data collected by different research groups


NCIT

“a biomedical vocabulary that provides consistent, unambiguous codes and definitions for concepts used in cancer research”

“exhibits ontology-like properties in its construction and use”.


Goals

to make use of current terminology “best practices” to relate relevant concepts to one another in a formal structure, so that computers as well as humans can use the Thesaurus for a variety of purposes, including the support of automatic reasoning;

to speed the introduction of new concepts and new relationships in response to the emerging needs of basic researchers, clinical trials, information services and other users.


Formal Definitions

of 37,261 nodes, 33,720 were stipulated to be primitive in the DL sense

Thus only a small portion of the NCIT ontology can be used for purposes of automatic classification and error-checking by using OWL.


Verbal Definitions

About half the NCIT terms are assigned verbal definitions

Unfortunately some are assigned more than one


Disease Progression

Definition1Cancer that continues to grow or spread.

Definition2 Increase in the size of a tumor or spread of cancer in the body.

Definition3 The worsening of a disease over time. This concept is most often used for chronic and incurable diseases where the stage of the disease is an important determinant of therapy and prognosis.


To make matters worse Disease Progression has as subclass:

Cancer Progression

Definition:

The worsening of a cancer over time. This concept is most often used for incurable cancers where the stage of the cancer is an important determinant of therapy and prognosis.


Cancer

a process (of getting better or worse)

an object (which can grow and spread)


Confuses definitions with descriptions

Tuberculosis DefinitionA chronic, recurrent infection caused by the bacterium Mycobacterium tuberculosis. Tuberculosis (TB) may affect almost any tissue or organ of the body with the lungs being the most common site of infection. The clinical stages of TB are primary or initial infection, latent or dormant infection, and recrudescent or adult-type TB. Ninety to 95% of primary TB infections may go unrecognized. Histopathologically, tissue lesions consist of granulomas which usually undergo central caseation necrosis. Local symptoms of TB vary according to the part affected; acute symptoms include hectic fever, sweats, and emaciation; serious complications include granulomatous erosion of pulmonary bronchi associated with hemoptysis. If untreated, progressive TB may be associated with a high degree of mortality. This infection is frequently observed in immunocompromised individuals with AIDS or a history of illicit IV drug use.


Confuses definitions with descriptions

Tuberculosis DefinitionA chronic, recurrent infection caused by the bacterium Mycobacterium tuberculosis. Tuberculosis (TB) may affect almost any tissue or organ of the body with the lungs being the most common site of infection. The clinical stages of TB are primary or initial infection, latent or dormant infection, and recrudescent or adult-type TB. Ninety to 95% of primary TB infections may go unrecognized. Histopathologically, tissue lesions consist of granulomas which usually undergo central caseation necrosis. Local symptoms of TB vary according to the part affected; acute symptoms include hectic fever, sweats, and emaciation; serious complications include granulomatous erosion of pulmonary bronchi associated with hemoptysis. If untreated, progressive TB may be associated with a high degree of mortality. This infection is frequently observed in immunocompromised individuals with AIDS or a history of illicit IV drug use.


A better definition

Tuberculosis

Definition:

A chronic, recurrent infection caused by the bacterium Mycobacterium tuberculosis.


NCIT inherits this ontological and terminological incoherence from source vocabularies in UMLS

Conceptual Entities =def

An organizational header for concepts representing mostly abstract entities.

Includes as subtypes:

action, change, color, death, event, fluid, injection, temperature


Conceptual Entities =def

An organizational header for concepts representing mostly abstract entities.

Confuses use and mention (swimming is healthy and has eight letters)


Duratec, Lactobutyrin, Stilbene Aldehyde

are classified by the NCIT as Unclassified Drugs and Chemicals


and problematic synonymsAnatomic Structure, System, or Substance ~ Anatomic

Structures and Systems

Does ‘anatomic’ apply only to structure or also to system and substance?

Biological Function ~ Biological Processsome biological processes are the exercises of biological

functionsothers (e.g. pathological processes, side effects) not

Genetic Abnormality ~ Molecular Abnormality (with subtype: Molecular Genetic Abnormality) (definitions not supplied)


Problematic synonyms

Diseases and Disorders ~ Disease ~ Disorder

Definition1 for Disease:A disease is any abnormal condition of the body or mind

that causes discomfort, dysfunction, or distress to the person affected or those in contact with the person. ...

Definition2 for DiseaseA definite pathologic process with a characteristic set of

signs and symptoms. ...

Condition ProcessDefinition2 contradicts NCIT’s own classification hierarchy


Three disjoint classes of plants

Vascular Plant

Non-vascular Plant

Other Plant


Three kinds of cells

Abnormal Cell is a top-level class (thus not subsumed by Cell

Normal Cell is a subclass of Microanatomy.

Cell is a subclass of Other Anatomic Concept (so that cells themselves are concepts)


NCIT as now constituted will block automatic reasoning

Neither Normal Cells nor Abnormal Cells are Cells within the context of the NCIT


Some consolations

NCIT is open source

NCIT has broad coverage

NCIT has some formal structure (OWL-DL)

NCIT is much, much better than (for example) the HL7-RIM

NCIT has realized the errors of its ways


The road ahead

http://www.cbd-net.com/index.php/search/show/938464

= “Review of NCI Thesaurus and Development of Plan to Achieve OBO Compliance”

and welcome to the Pre-NCIT:http://nciterms.nci.nih.gov/NCIBrowser/Dictionary.do




Fragment of Pre-NCIT Hierarchy

Murine Tissue Type Body Fluids and Substances (MMHCC) Cardiovascular System (MMHCC) Blood Vessel (MMHCC) Heart (MMHCC) Digestive System (MMHCC)


First step

Alignment of OBO ontologies through a common system of formally defined relations in the OBO-RO (OBO Relation Ontology)

see “Relations in Biomedical Ontologies”, Genome Biology Apr. 2005


is_a (sensu UMLS)

A is_a B =def

‘A’ is narrower in meaning than ‘B’

grows out of the heritage of dictionaries

(which ignore the basic distinction between types and instances)


To build a high quality shared ontology requires hard work and

staying power

You cannot cheat by borrowing from UMLS

UMLS (= the UMLS Metathesaurus) is not an ontology


Concepts, Concept Names, and their Identifiers in the UMLS

The Metathesaurus is organized by concept. One of its primary purposes is to connect different names for the same concept from many different vocabularies.

A concept is a meaning. A meaning can have many different names. A key goal of Metathesaurus construction is to understand the intended meaning of each name in each source vocabulary and to link all the names from all of the source vocabularies that mean the same thing (the synonyms). This is not an exact science. ... Metathesaurus editors decide what view of synonymy to represent in the Metathesaurus concept structure. Please note that each source vocabulary’s view of synonymy is also present in the Metathesaurus, irrespective of whether it agrees or disagrees with the Metathesaurus view.


This strange mapping

between names as they appear in different source vocabularies created for widely different purposes can still be very useful

but the source vocabularies themselves are of variable quality

(not all mappings are created equal)and the sorts of search which the UMLS

supports reflects an already outmoded technology


is_a

congenital absent nipple is_a nipple

surgical procedure not carried out because of patient’s decision is_a surgical procedure

cancer documentation is_a cancer

disease prevention is_a disease

living subject is_a information object representing an animal or complex organism

individual allele is_a act of observation

limb is_a tissue


is_a (sensu UMLS)

both testes is_a testis

plant leaves is_a plant

smoking is_a individual behavior

walking is_a social behavior


is_a

A is_a B =def

For all x, if x instance_of A then x instance_of B

cell division is_a biological process

adult is_a child ???


Two kinds of entities

occurrents (processes, events, happenings)

cell division, ovulation, death

continuants (objects, qualities, ...)

cell, ovum, organism, temperature of organism, ...


is_a (for occurrents)

A is_a B =def

For all x, if x instance_of A then x instance_of B

cell division is_a biological process


is_a (for continuants)

A is_a B =def

For all x, t if x instance_of A at t then x instance_of B at t

abnormal cell is_a celladult human is_a humanbut not: adult is_a child


part_of

Composes, with one or more other physical units, some larger whole.

(UMLS Semantic Network)

what does this relation relate?

A is_a B =def A is narrower in meaning than B


Part_of as a relation between types is more problematic than

is standardly supposed

heart part_of human being ?

human heart part_of human being ?

human being has_part human testis ?

testis part_of human being ?


Definition of part_of as a relation between types

A part_of B =Def all instances of A are instance-level parts of some instance of B

human testis part_of adult human being


two kinds of parthood

1. between instances:

Mary’s heart part_of Mary

this nucleus part_of this cell

2. between types

human heart part_of human

cell nucleus part_of cell


part_of (for occurrents)

A part_of B =def.

For all x, if x instance_of A then there is some y, y instance_of B and x part_of y

where ‘part_of’ is the instance-level part relation

EVERY A IS PART OF SOME B


part_of (for continuants)

A part_of B =def.

For all x, t if x instance_of A at t then there is some y, y instance_of B at t and x part_of y

where ‘part_of’ is the instance-level part relation

NOTE THE ALL-SOME STRUCTURE


A part_of B, B part_of C ...

The all-some structure of such definitions allows

cascading of inferences

(i) within ontologies

(ii) between ontologies

(iii) between ontologies and EHR repositories of instance-data


Cascading inferences

Whichever A you choose, the instance of B of which it is a part will be included in some C, which will include as part also the A with which you began

The same principle applies to the other relations in the OBO-RO:

located_at, transformation_of, derived_from, adjacent_to, etc.


is_a and part_of never cross categorial divides

(cf. tripartite organization of GO)

if A is_a B

then A is an object type iff B is an object type

then A is a process type iff B is a process type

then A is a characteristic type iff B is a characteristic type


Kinds of relations

Between types:– is_a, part_of, ...

Between an instance and a type– this explosion instance_of the type explosion

Between instances:– Mary’s heart part_of Mary


Continuityinstance a continuous_with instance b

is always symmetric

But consider the types lymph node and lymphatic vessel:

Each lymph node is continuous with some lymphatic vessel, but there are lymphatic vessels (e.g. lymphs and lymphatic trunks) which are not continuous with any lymph nodes

Continuity on the type level is not symmetric.


Adjacency as a relation between universals is not

symmetric

Consider

seminal vesicle adjacent_to urinary bladder

Not: urinary bladder adjacent_to seminal vesicle


Instance level

this nucleus is adjacent to this cytoplasm

implies:

this cytoplasm is adjacent to this nucleus

Type level

nucleus adjacent_to cytoplasm

Not: cytoplasm adjacent_to nucleus


Applications

Expectations of symmetry e.g. for protein-protein interactions hmay hold only at the instance level

if A interacts with B, it does not follow that B interacts with A

if A is expressed simultaneously with B, it does not follow that B is expressed simultaneously with A


Definitions of the all-some form

allow cascading inferences

If A R1 B and B R2 C, then we know that

every A stands in R1 to some B, but we know also that, whichever B this is, it can be plugged into the R2 relation


GALEN: Vomitus contains carrot

All portions of vomit contain all portions of carrot

All portions of vomit contain some portion of carrot

Some portions of vomit contain some portion of carrot

Some portions of vomit contain all portions of carrot


c at t1

C

c at t

C1

time

same instance

transformation_of

pre-RNA mature RNA

adultchild


transformation_of

A transformation_of B =Def. Every instance of A was at some earlier time an

instance of B

adult transformation_of child


embryological development C

c at t c at t1

C1


C

c at t c at t1

C1

tumor development


C

c at t

C1

c1 at t1

C'

c' at t

time

instances

zygote derives_fromovumsperm

derives_from


Request from Bill Bug

How best to effectively bring together:- spatial/morphological ontologies; - neuroscience terminologies (e.g.,

NeuroNames) and; - data-centric neuroanatomical indexing

systems (voxel-based 3D atlases);to promote optimal integration of neuroscience data sets that include a spatial component, however coarse.


A suite of defined relations between universals

Foundational is_apart_of

Spatial located_incontained_inadjacent_to

Temporal transformation_ofderives_frompreceded_by

Participation has_participanthas_agent


Logical Theory of Spatial Relations

RCC: Region-Connection Calculus (Leeds Qualitative Spatial Reasoning Group)

Cf. Dameron et al. Modeling dependencies between relations to ensure consistency of a cerebral cortex anatomy knowledge base


Principles

1 anatomical structure 1 regionhas_location

Define the relationships of adjacency, connectedness etc. using RCC-8 and its extensions

DC EC PO TPP NTPP EQ


Example 1Reasoning with part and location at the

instance level:

Inferior Frontal Gyrus Operc. Pars of Inferior Frontal Gyrus

Frontal Lobe


Example 2

Reasoning with location, continuity and external connection

PreCentral Gyrus PostCentral GyrusFrontal Lobe


Extension to the 3-D case

x

y

substances x, y participate in process B

time

Bx

y

SNAP-ti.

time

SPAN

B

slice of x’s life


Most ontologies are execrableBut some good ontologies do already

exist

• as far as possible don’t reinvent• use the power of combination and collaboration• ontologies are like telephones: they are valuable

only to the degree that they are used and networked with other ontologies

• but choose working telephones• most UMLS telephones were broken from the

start


Why do we need rules/standards for good ontology?

Ontologies must be intelligible both to humans (for annotation) and to machines (for reasoning and error-checking): unintuitive rules for classification lead to errors

Intuitive rule facilitate training of curators and annotators

Common rules allow alignment with other ontologies

Logically coherent rules enhance harvesting of content through automatic reasoning systems


To the degree that basic rules of good ontology are not satisfied, error checking and ontology alignment will be achievable, at best, only– with human intervention – via force majeure– with unstable results


Current practice in the domain of clinical research

Results of clinical trials are organized too tightly around specific diagnostic criteria imposed by specific, local, hypotheses

A change in criteria forces a costly re-examination and re-coding of all existing records to make them usable in future hypothesis generation and testing.


How to solve this problem?

Just as clinical hypotheses need to be tied to basic science, so special-purpose application ontologies need to be tied to general-purpose reference ontologies


We separate data as interpreted in terms of current criteria

from the structure of the underlying biomedical reality

and ensure that the first is stored and processed always by using terms drawn from a shared, stable representation (a reference ontology) of the latter.

Diagnostic criteria for a disease can then be changed but we will still maintain access to the data relevant to all prior diagnosed cases of the disease in question.

How to solve this problem?


Not only data needs to be aligned through pre-established reference ontologies, so also does softwareCurrently, application ontologies are built afresh for each new application

They commonly introduce new idiosyncrasies of terminology, format or logic, plus simplifications or distortions of their subject-matters.

This may do no harm in relation to the specific application (for example radiology, tissue classification, cancer staging) – and keeps the software simple


But what happens

when other applications want to use the data annotated in their terms, or when we need to extend to a larger portion of biomedical reality?Now the expanded ontology will no longer be compatible with the software designed for its original application. Different groups now need to start working with different and incompatible versions of an ontology, engendering a spiralling complexity as these different versions themselves become revised and extended for different purposes.


The solution

The methodology of always developing application ontologies against the backgrund of formally robust reference ontologies can both counteract these tendencies toward ontology proliferation and ensure the interoperability of application ontologies as they become further developed in the future.


The methodology of reference ontologies

can provide locally developed application ontologies with cross-granular understanding of the ways processes at the gene and protein level are linked to clinically salient processes at coarser granularity

and it can allow them take advantage of existing logical tools and methods for reasoning across large bodies of data.


An application ontology

is comparable to an engineering artifact such as a software tool. It is constructed for a specific practical purpose.

Examples:

NCIT

FuGO Functional Genomics Investigation Ontology


A reference ontology

A reference ontology has a unified subject-matter, which consists of entities existing independently of the ontology, and it seeks to optimize descriptive or representational adequacy to this subject matter.

A reference ontology is analogous to a scientific theory. Thus it consists of representations of biological reality which are correct when viewed in light of our current understanding of reality, and it must be subjected to updating in light of scientific advance.

Example: The Foundational Model of Anatomy


Current Best Practice



Pleural Cavity

Pleural Cavity

Interlobar recess

Interlobar recess

Mesothelium of Pleura


Pleura(Wall of Sac)

Pleura(Wall of Sac)

VisceralPleura

VisceralPleura

Pleural SacPleural Sac

Parietal Pleura

Parietal Pleura

Anatomical SpaceAnatomical Space

OrganCavityOrganCavity

Serous SacCavity

Serous SacCavity

AnatomicalStructure

AnatomicalStructure

OrganOrgan

Serous SacSerous Sac

MediastinalPleura

MediastinalPleura

TissueTissue

Organ PartOrgan Part

Organ Subdivision

Organ Subdivision

Organ Component

Organ Component

Organ CavitySubdivision


Serous SacCavity

Subdivision

Serous SacCavity

Subdivision

part

_of

is_a


The Foundational Model of Anatomy

Follows formal rules for ‘Aristotelian’ definitions

When A is_a B, the definition of ‘A’ takes the form:

an A =def. a B which ...

a human being =def. an animal which is rational


FMA Example

Cell =def. an anatomical structure which consists of cytoplasm surrounded by a plasma membrane with or without a cell nucleus

Plasma membrane =def. a cell part that surrounds the cytoplasm


The FMA regimentation

Brings the advantage that each definition reflects the position in the hierarchy to which a defined term belongs.

The position of a term within the hierarchy enriches its own definition by incorporating automatically the definitions of all the terms above it.

The entire information content of the FMA’s term hierarchy can be translated very cleanly into a computer representation


GO now adopting structured definitions which contain both genus and differentiae

Species =def Genus + Differentiae

neuron cell differentiation =defdifferentiation by which a cell acquires features of a neuron


Ontology alignmentOne of the current goals of GO is to align:

cone cell fate commitment retinal_cone_cell

keratinocyte differentiation keratinocyte

adipocyte differentiation fat_cell

dendritic cell activation dendritic_cell

lymphocyte proliferation lymphocyte

T-cell homeostasis T_lymphocyte

garland cell differentiation garland_cell

heterocyst cell differentiation heterocyst

Cell Types in GO Cell Types in the Cell Ontologywith


Alignment of the two ontologies will permit the generation of consistent and complete definitions

id: CL:0000062name: osteoblastdef: "A bone-forming cell which secretes an extracellular matrix. Hydroxyapatite crystals are then deposited into the matrix to form bone." [MESH:A.11.329.629]is_a: CL:0000055relationship: develops_from CL:0000008relationship: develops_from CL:0000375

GO

Cell type

New Definition

+

=Osteoblast differentiation: Processes whereby an osteoprogenitor cell or a cranial neural crest cell acquires the specialized features of an osteoblast, a bone-forming cell which secretes extracellular matrix.


Other Ontologies to be aligned with GO

Chemical ontologies– 3,4-dihydroxy-2-butanone-4-phosphate synthase

activity

Anatomy ontologies– metanephros development

GO itself– mitochondrial inner membrane peptidase activity

OBO core


eventually to comprehend all of OBO


Pleural Cavity

Pleural Cavity

Interlobar recess

Interlobar recess



Pleura(Wall of Sac)

Pleura(Wall of Sac)

VisceralPleura

VisceralPleura


Parietal Pleura

Parietal Pleura



Serous SacCavity

Serous SacCavity

AnatomicalStructure

AnatomicalStructure

OrganOrgan


MediastinalPleura

MediastinalPleura

TissueTissue


Organ Subdivision

Organ Subdivision

Organ Component

Organ Component



Serous SacCavity

Subdivision

Serous SacCavity

Subdivision

part

_of

is_a


Anatomical Entity

Physical Anatomical Entity

Material Physical Anatomical Entity

-is a-

Non-material Physical Anatomical Entity

ConceptualAnatomical Entity

AnatomicalStructure

BodySubstance

BodyPart

HumanBody

OrganSystem

OrganCell

OrganPart

AnatomicalSpace

Anatomical Relationship

CellPart

Biological Macromolecule

Tissue


The Anatomy Reference Ontology

is organized in a graph-theoretical structure involving two sorts of links or edges:

is-a (= is a subtype of )

(pleural sac is-a serous sac)

part-of

(cervical vertebra part-of vertebral column)


at every level of granularity


What do the kidneys do?Modularity


How does a kidney work?NEPHRON


Nephron FunctionsFUNCTIONAL SEGMENTS


Top-Level Categories in the FMAanatomical

entity

non-physicalanatomical entity

physicalanatomical entity

anatomical relationship

body substance

material physical anatomical entity

anatomical structure

non-material physical anatomical entity

body space

boundary anatomical attribute


anatomical structure (cell, lung, nerve, tooth)

result from the coordinated expression of structural genes

have their own 3-D shape


portion of body substance

inherits its shape from container

portion of urine

portion of menstrual fluid

portion of blood


anatomical space

cavities, conduits


anatomical attribute

mass

weight

temperature

your temperature

its value now


anatomical relationship

located_in

contained_in

adjacent_to

connected_to

surrounds

lateral_to (West_of)

anterior_to


boundary

bona fide / fiat

www.enel.ucalgary.ca/ People/Mintchev/stomach.htm

http://www.enel.ucalgary.ca/People/Mintchev/stomach.htm




Connectedness and Continuity

The body is a highly connected entity. Exceptions: cells floating free in blood

continuous_with, attached_to (muscle to bone) synapsed_with (nerve to nerve and nerve

to muscle)Two continuants are continuous on the instance

level if and only if they share a fiat boundary.


Pleural Cavity

Pleural Cavity

Interlobar recess

Interlobar recess



Pleura(Wall of Sac)

Pleura(Wall of Sac)

VisceralPleura

VisceralPleura


Parietal Pleura

Parietal Pleura



Serous SacCavity

Serous SacCavity

AnatomicalStructure

AnatomicalStructure

OrganOrgan


MediastinalPleura

MediastinalPleura

TissueTissue


Organ Subdivision

Organ Subdivision

Organ Component

Organ Component



Serous SacCavity

Subdivision

Serous SacCavity

Subdivisionbasis for generalization to other species


Pleural Cavity

Pleural Cavity

Interlobar recess

Interlobar recess



Pleura(Wall of Sac)

Pleura(Wall of Sac)

VisceralPleura

VisceralPleura


Parietal Pleura

Parietal Pleura



Serous SacCavity

Serous SacCavity

AnatomicalStructure

AnatomicalStructure

OrganOrgan


MediastinalPleura

MediastinalPleura

TissueTissue


Organ Subdivision

Organ Subdivision

Organ Component

Organ Component



Serous SacCavity

Subdivision

Serous SacCavity

Subdivision

part

_of

is_a


Web-Based Representations of Neuroanatomy



includes Neuronames


Human Morphometry and Function BIRN Testbeds

with thanks to Christine Fennema-Notestine and Jessica Turner

CBiO/BIRN Workshop 2006


BIRN Ontology NeedsGOAL: User will employ BIRN interface and Mediator

to perform scientific queries on data from• structural and functional MRI experiments• clinical assessments• psychiatric interviews• and/or behavioral experiments

BIRN needs for common vocabularies– Mediator needs to talk across databases to find

relevant/similar information; this requires linking of concepts to table columns and values

– Query interface needs semantic network to find related information


Example queries:

– Find all datasets of schizophrenics with structural and functional imaging data related to working memory

– Find the correlation between hippocampal volume and working memory performance in AD subjects


MBIRN priorities

“To relate clinical assessments, cognitive function, and neuroanatomy within mBIRN’s multi-site AD sample, with future branching into neuropsychiatric measures”

– Only a high quality reference ontology of neuro(patho)anatomy from the macroscopic to the subcellular levels of granularity can give you this


Existing neuroanatomical ontology

Need to create related “function”-based ontology

Brain

Cerebellum Cerebrum

Cerebral white matter …

Frontal cortex Temporal cortex

Superior temporal Mesial temporal

Amygdala Hippocampus

…

Cerebral cortex

…

…

…

Memory

CVLT


‘Need to create related “function”-based ontology’

UMLS: mental process is_a organism function

Function vs. functioning

Many entities have functions which they never realise

A has function B = A can realise B (under which circumstances?)



A function is a disposition of an independent continuant to engage in corresponding processes.

To what extent are the various functions identified by BIRN are in fact complex processes with many less complex processes as their parts.

How are functions different from disfunctions / malfunctions ?

Are all function such that their execution is good for the organism?



“You cannot classify parts of the brain on the basis of which parts can think, remember, effect movement or perceive various kinds of sensations, just as you cannot sort anatomical entities on the basis of which can pump, digest, secrete, fertilize or stabilize.”

“It is impossible to create an inheritance class subsumption hierarchy of neuroanatomical entities at any meaningful depth on the basis of function.”


Brain

Cerebrum

Temporal

Mesial temporal

Hippocampus

Cerebral cortex

CVLT

Task and score description

Frontal Cognitiveimpairment

Cognition

Assessment

Neuropsychology

Amnesia

Memory Learning


Memory

CVLT SIRP

Assessment

Behavioral Paradigm

Cognitive Process

Attention

Working memory Long Term memory

SCID-Patient

Breathhold

Action

Can we reason on the basis of a graph of this sort?


Bonfire Relations

relation: the type of relation between the concept to the left and the concept to the rightPAR = ParentCHD = ChildSIB = SiblingRB = Broader RelationshipRN = Narrower RelationshipRO = Other Relationship


BIRN Relations

UMLS (PAR, CHD, RN, RO, RB, SY).RB: has a broader relationship RN: has a narrower relationship RO: has relationship other than

synonymous, narrower, or broader CHD: has child relationship in a

Metathesaurus SIB: has sibling relationship in a

Metathesaurus source vocabulary


“Circular Hierarchical Relationships in the UMLS:Etiology, Diagnosis, Treatment, Complications and Prevention”

Olivier Bodenreider

Topographic regions: General terms

Physical anatomical entity

Anatomical spatial entity

Anatomical surface

Body regions

Topographic regions


MeSH

MeSH Descriptors Index Medicus Descriptor Anthropology, Education, Sociology and Social Phenomena (MeSH Category) Social Sciences Political Systems National Socialism

National Socialism is_a Political SystemsNational Socialism is_a Anthropology ...


MeSH

National Socialism is_a MeSH Descriptor

Cf. NeuroNames:

Ontology =def a codification of the relationships between words and concepts


Human BIRN data includes:Participant demographics such as age, gender, …

Clinical and psychiatric information – Assessments used, data type– Diagnostic information

Behavioral data during fMRI tasks– Need to know how to interpret that (“is a button 1 response

a yes or a no?”)

Raw structural and functional images– Need information about data collection and preprocessing

methods

Single-subject and group level analyses and results– Need information about analytic methods used


Areas where application ontologies will be needed

Participant demographics such as age, gender, …

Clinical and psychiatric information – Assessments used, data type– Diagnostic information

Behavioral data during fMRI tasks– Need to know how to interpret that (“is a button 1 response

a yes or a no?”)

Raw structural and functional images– Need information about data collection and preprocessing

methods

Single-subject and group level analyses and results– Need information about analytic methods used


Bottom-up search:User’s dataset contains the CVLT – what does it measure?

• Search for CVLT• Related to PARENT concepts like “Neuropsychological

tests” or “Assessment Scales” or SIBLING concepts of other tests

• What is the CVLT? This doesn’t answer the user’s question.

• Need relationship links to function: memory and learning

• Need relationship links to structure: anatomical regions reflected in change of performance on this measure hippocampus


Top-down search:

User interested in studying the relationship between hippocampal volume and memory performance in Alzheimer’s disease.• Search for measures of memory• Would like to see memory linked to CVLT • Would like to see memory linked to hippocampus at a

very basic level • Would like to see links to potential disorders assessed,

e.g., amnesia or AD


Ontology/Terminology Infrastructure

GOAL: to allow database mediation and scientific queries for multi-site clinical neuroimaging studies. This requires the relationship of database tables to concepts and to relate brain structure and function through neuroanatomical regions, neuropsychological and cognitive terms, and clinical assessments.



– To do this, the Mediator relies in part on defined terms/concepts to define relationships between similar terms from different databases.

– If a user is interested in data related to “long delay free recall," it is important to also include information related to “memory." This type of relational knowledge is critical to find other values in other databases that have similar information.



In addition, the ontology will provide a semantic network; for a user searching for “memory" information, related information would include

– Cognitive terms, e.g., recall, recognition, short and long term memory

– Assessment terms, e.g., California Verbal Learning Test

– “Disorders of” terms, e.g., Alzheimer’s disease is a disorder of memory

How block information overload?


Bottom-up search:User’s resultant dataset contains the MMSE – the user asks what does it measure?• Search for MMSE concept• Related to PARENT concepts like Neuropsychological tests” or

“Assessment Scales” or SIBLING concepts of other tests • What is the MMSE? This doesn’t answer the user’s question.• Need relationship links to function: general cognitive ability,

cognitive impairment, dementia severity, brain damage …• Need relationship links to structure: anatomical regions reflected in

change of performance on this measure, although a relatively non-specific measure


Top-down search:What variables exist that would provide a measure

of general cognitive function and dementia severity?• Search for measures of (general) cognitive function• Would like to see general cognitive ability, cognitive

impairment, dementia severity linked to MMSE • Would like to see general cognitive ability, cognitive

impairment, dementia severity linked to neuroanatomical regions, simply brain in this case

• Would like to see links to potential disorders measured, e.g., AD


NeuroNames (with thanks to Onard Mejino)

has a limited scope.

It deals with neuroanatomical structures only at the gross level. No cellular, subcellular or macromolecular entities are represented.

The peripheral nervous system and the spinal cord are not included.

It represents structures from different species (human, macaque and rodent) in the same hierarchy.


NN’s main hierarchyis a partonomy based on mutually exclusive and exhaustive volumetric partitions, the equivalent of regional partition in the FMA. The partonomy supports only ONE partition view and therefore does not accommodate

• other recognized regional partitions like Brodman areas (treated as “ancillary structures”)

• constitutional parts like the internal pyramidal layer of neocortex and the vasculature of neuraxis (entities that have important clinical significance)

• new partitions advanced by new technology like gene expression mappings or radiologic imaging techniques

• partitions determined by formal spatial region-based ontologies like RCC


The Neuronames partonomy

will serve at best as an application ontology for annotating segmented images of the brain. But it will still be very difficult to link the annotated image data to all the other types of data which will BIRN will need to describe

a reference ontology of neuroanatomy is a first priority.


Neuronames

• Since univocity is not enforced in the literature of neuroanatomy, e.g. the term ‘Basal ganglia’ represents different structures when used in association with anatomic, functional and clinical views.

• How will NN resolve or clarify this?


Neuronames

• entities are primarily identified on the basis of stains that distinguish gray matter from white matter

• thus not on principles or rules that define the type of the entity in question, thereby yielding a partition not in accord with the standards commonly accepted for representing the rest of the body.

• gray matter and white matter are viewed as tissues. But tissue is usually defined as an aggregate of similarly specialized cells and intercellular matrix.

• yet gray matter consists not of cells but of cell bodies, white matter not of cells but of neurites


Neuronames

• gives no explicit definitions, and the representations it gives (e.g. of the Fourth Ventricle*) are often at odds with consensual usage

• hence scalability, extendability, combinability with other ontologies is limited – how then can it be used to bridge research efforts at the genomic / proteomic level with those at the clinical level?

• Information unique to neuroanatomical entities such as axonal input/output relationships, connectivity, neuron type, neurotransmitter and receptor types are indispensable in establishing and understanding both structural and physiological relationships among neuroanatomical entities and their relationship with the rest of the body.


BIRNLex

does provide definitions, normally taken over from UMLS


Rules for definitions

‘A’ = child term‘B’ = parent term

an A =def a B which Cs

If a definition is correct it should always make sense to substitute ‘a B which Cs’ for ‘an A’

“A human being is subject to processes of aging”“A rational animal is subject to processes of aging”


BIRNLex

The eye =def.

The eyeball and its constituent parts, e.g. retina

mouse =def.

common name for the species mus musculus


BIRNLex


BIRNLex


BIRNLex


BIRNLex

bear in mind always that your ontology needs

to be interoperable with other ontologies


BIRNLex

bear in mind always that your ontology needs

to be interoperable with other ontologies


BIRNLex

surface =def 3D segmentation obtained by fitting a polygonal mesh around the boundary of an object of interest, creating a 3D surface

Concept =def Generic ideas or categories derived from common properties of objects, events, or qualities, usually represented by words or symbols


BIRNLex

brain imaging =def none; synonymous with positrocephalogram, nos

CA1 =def CA1 cytoarchitectonic field of hippocampus

cognitive process = def. conceptual function or thinking in all its forms


BIRNLex and UMLS-SN

Rest =SN Daily or Recreational ActivityPrincipal Investigator =SN Professional or Occupational Group

Left handedness =SN Organism AttributeAmbidextrous =SN Finding

Brain Imaging =SN Diagnostic ProcedureBrain Mapping =SN Diagnostic Procedure & Research Activity

Healthy Adult =SN Finding


BIRNLex

Mouse BIRN: Ontologies

Maryann Martone

and

Bill Bug

2005 All Hands Meeting

Mouse BIRN: Ontologies

Maryann Martone and Bill Bug


Use of Ontologies in BIRN•Databases

•Enforces semantic consistency within a database

•Data Sharing•Establishes semantic relationship among concepts contained in distributed databases

•Data integration•Bridging across multiscale and multimodal data

•Concept-based queries:•Ontologies can be used to alter semantic context to present a view of the conceptual aspects of a data set or meta-analysis result most relevant to a particular neuroscientist


Objectives of Working Group

Educate BIRN participants on the use of ontologies and associated tools for data integration– Tuesday (PM) and Wednesday (AM)

Develop a set of ontology resources for BIRN participants, based on existing ontologies where possible

Identify areas that are not well covered by existing ontologies for possible development.

***Develop a clear set of policies and procedures for working with ontologies– Including curation, addition of core ontologies, extension of

ontologies, mapping of databases to ontologies


Goals of OTF

•Provide a dynamic knowledge infrastructure to support integration and analysis of BIRN federated data sets, one which is conducive to accepting novel data from researchers to include in this analysis.

•Identify and assess existing ontologies and terminologies for summarizing, comparing, merging, and mining datasets. Relevant subject domains include clinical assessments, demographics, cognitive task descriptions, imaging parameters/data provenance in general, and derived (fMRI) data.

•Identify the resources needed to achieve the ontological objectives of individual test-beds and of the BIRN overall. May include finding other funding sources, making connections with industry and other consortia facing similar issues, and planning a strategy to acquire the necessary resources.


BIRN Ontology Resources

Mouse BIRN Ontology Resource Page

http://nbirn.net/Resources/Users/Ontologies/

Bonfire Ontology Browser and Extension Tool


Current Ontology Development by Mouse BIRN Participants

Developmental Ontology• Seth Ruffins, Cal Tech

Subcellular Anatomy• Maryann Martone and Lisa Fong, UCSD


Ontology for Subcellular Anatomy of Nervous System


CCDB DictionaryTerm Ontology ConceptID Semantic Type Definition

Cerebellum UMLS C0007765 Body Part, Organ, or Organ Component

Part of the metencephalon that lies in the posterior cranial fossa behind the brain stem. It is concerned with the coordination of movement. (MSH)

Glial Fibrillary Acidic Protein

UMLS C0017626 Amino Acid, Peptide, or Protein, Biologically Active Substance

An intermediate filament protein found only in glial cells or cells of glial origin. MW 51,000. (MSH)

Medium Spiny Neuron

Bonfire BID000012 Cell Small (10-15 µm in diameter) projection neurons found in neostriatum, possessing a rougly spherical dendritic tree composed of 3-5 primary dendrites. Dendrites are covered with dendritic spines.

Purkinje cell UMLS C0034143 Cell large branching neurons of the middle layer of cerebellar cortex, characterized by vast arrays of dendrites; the output neurons of the cerebellar cortex.


Some Areas of Interest to BIRN

Navigating through Multi-resolution information

Linking animal and human imaging data

brain

cerebellum

cerebellar cortex

Purkinje cell

dendritic spine

Entopeduncular nucleus

Globus pallidus, internal segment

Animal Model Disease Process

•***Map between Human and Animal models

•Functional assessment


Anatomical Knowledge SourcesFoundational model of anatomyNeuronames (Brain Info)***BAMS***Adult Mouse Anatomical Dictionary

(Edinburgh/GO)

“Although BIRN is an open, diverse and fluid environment, the use of ontologies for enhanced interoperability will be pointless if we allow random use of ontologies. The OTF recommends that there be a set of ontologies that are approved for use and a set of policies and procedures for adding or creating additional knowledge sources. Current knowledge sources that are currently in use include UMLS, GO, LOINC, SNOMED, NEURONAMES.”

-OTF report to BEC 8/05


Other Resources Likely of Use

Mouse Phenome Project: a collection of phenotypic and genotypic data for the laboratory mouse

anatomybehaviorbiological factorsbloodcancerdiet effectsdrug effects, toxicitygenotype heart, lung intake, metabolism musculoskeletal neurosensory reproduction


Neuronames-UMLS-Smart Atlas

•Mapping of rodent nomenclature onto UMLS

•Neuronames has now included many of the terms

•Using concepts in Neuronames and Paxinos to create new hierarchy


What do we need to do in the next year

Identify areas of mouse BIRN not covered– Do ontologies exist?– If not, do we develop them

What known ontologies should be added to BIRN ontology resources– Who will handle semantic concordance– How do we represent these in BIRN?

Mapping databases to ontologies– Time frame– What should be mapped?– Who will do this at each site


Mouse BIRN Global Conceptual Schema

Project

ExperimentalData

MolecularDistributions

Subject

Animal Type

Experiments

AnatomicalProperties

MicroarrayResults

Images

Atlas

Region ofInterest

Worked with Data Integration group to define global schema

Documents

10:30-12:00 How to Build an Ontology 1-2pm Best Practices and Lessons Learned 2-3pm BIRN Ontologies: An Overview