The NCI Thesaurus,

1
The NCI Thesaurus, Overview and Utilization
Gilberto Fragoso NCI Center for
Bioinformatics SOFG2, Oct 26, 2004
2
Enterprise Vocabulary Services

• Services and resources that address NCI's needs
  for controlled vocabulary (http//ncicb.nci.nih.go
  v/core/EVS)
• An NCI collaboration
• NCI Office of Communications
• Cancer Information Products and Systems
• PDQ and Cancer.gov
• NCI Center for Bioinformatics
• caCORE
• Community portals
• Vocabulary Products
• NCI Thesaurus standalone terminology
• NCI Metathesaurus maps vocabularies
• External vocabularies

3
NCI Thesaurus

• Reference Terminology for NCI
• Broad coverage of cancer domain
• Findings and Disorders
• Anatomy
• Drugs, chemicals
• Oncogenes, gene products, biological processes
• Cancer models - murine
• Research techniques, management

4
NCI Thesaurus - Content

• 38,000 Concepts (115,000 terms)
• Partitioned in 20 Kinds ( disjoint classes)
• Hierarchically organized
• Polyhierarchy but only within kinds
• e.g. Neoplasms in the Findings_and_Disorders_Kind
• Properties state facts about a Concept
• Clarify meaning of concepts
• Application support
• Roles, binary relations between Concepts

5
(No Transcript)
6
NCI Thesaurus - Content

• 38,000 Concepts (115,000 terms)
• Partitioned in 20 Kinds ( disjoint classes)
• Hierarchically organized
• Polyhierarchy but only within kinds
• e.g. Neoplasms in the Findings_and_Disorders_Kind
• Properties state facts about a Concept
• Clarify meaning of concepts
• Application support
• Roles, binary relations between Concepts

7
NCI Thesaurus - Properties
8
NCI Thesaurus - Roles

• Roles, binary relations between Concepts
• Utilized by the DL classifier to determine
  subsumption
• Used to express semantic relations
• Role selection in the NCI Thesaurus
• Define the minimal number of Roles that will
  allow concepts in a Kind to be declared defined
• Specific roles to support dependent applications
• Domain knowledge e.g. Diseases

9
Roles in the Disease Domain
10
Roles in the NCI Thesaurus
ftp//ftp1.nci.nih.gov/pub/cacore/EVS/ThesaurusSem
antics/Tbox04Mar.png
11
Various Kinds in the Thesaurus
ftp//ftp1.nci.nih.gov/pub/cacore/EVS/ThesaurusSem
antics/KindDefinitions.pdf
12
Sample Disease Concept
Gastric Mucosa-Associated Lymphoid Tissue
Lymphoma   Molecular abnormalities Disease_May_H
ave_Cytogenetic_Abnormality Trisomy
3 Disease_May_Have_Cytogenetic_Abnormality
Trisomy 18 Role group 1 Disease_May_Have_Cytog
enetic_Abnormality t(1118)(q21q21) Disease_Ma
y_Have_Molecular_Abnormality AP12-MLT fusion
protein expression Histogenesis Disease_Has_Norm
al_Cell_Origin Post-germinal center marginal
zone B-lymphocyte   Pathology Disease_Has_Abnorm
al_Cell Centrocyte-like cell Disease_May_Have_Ab
normal_Cell Neoplastic monocytoid
B-lymphocyte Disease_May_Have_Abnormal_Cell
Neoplastic plasma cell Disease_May_Have_Finding
Lymphoepithelial lesion   Anatomy Disease_Has_Pr
imary_Anatomic_Site Stomach Disease_Has_Normal_T
issue_Origin Gut associated lymphoid
tissue   Clinical information Disease_May_Have_F
inding Indolent clinical course Disease_May_Have
_Associated_Disease Hepatitis C
13
Retrieval of Indexed Documents
C1, C2, .., CX
User
Concepts used for retrieval
Search Engine
D1ltC1, C2gt
D2ltC1, C3, C4 gt
Relevant documents
Document
Indexing terms
14
History-Enhanced Retrieval of Indexed Documents
Thesaurus version
pre-indexed documents
R1
Version 1
new
R2
Version 2
modify
R3
merge
Version 3
retire
split
R4
Version 4
Traverse the history net to find the right
indexing terms for retrieval
Search Engine
Concepts used for retrieval
15
Representation of Edit Actions in History Table
16
caCORE Infrastructure
- Vocabulary Services - NCI Thesarus - NCI
Metathesaurus - caBIO API access
http//ncicb.nci.nih.gov/core
17
caCORE Infrastructure

• - Standard object models
• Uniform programmatic
• interface access

18
caBIO Architecture
19
caCORE Infrastructure

• - ISO 11179
• Metadata Registry
• CDEs
• UML Models
• Utilizes EVS
• caBIO API access

20
(No Transcript)
21
caCORE Infrastructure
CTEP DCP CIP SPORES
caImage caMod caArray
22
WebTree App in caImage
23
NCI Thesaurus Distribution Access

• Programmatic access via caBIO API
• Web browsable (http//nciterms.nci.nih.gov)
• Various file formats (ftp//ftp1.nci.nih.gov/pub/c
  acore/EVS/)
• Flat
• Ontylog xml (native)
• OWL Lite

24
Terminology Development in Protégé/OWL?

• - Merge
• Split
• Retirement
• History
• Codes
• Search
• Workflow

25
Additional Integration?

• Publication of ontologies in OWL
• Terminology
• Chute Solbrig (Mayo pers. comm.)
• Concept History

26
EVS Team

• NCI Office of Communications
• Margaret Haber
• Larry Wright
• NCI Center for Bioinformatics
• Frank Hartel
• Sherri de Coronado
• Gilberto Fragoso
• Ken Buetow, NCICB Director
• Peter Covitz, caCORE Director
• Apelon, Inc. Northrup Grumman, Inc.
• Aspen, Inc. Kevric Corporation
• SAIC J. Oberthaler Consulting
• Protégé/SMI