MULTIDISCIPLINRY NATURE OF THE ENVIRONMENTAL STUDIES.pptx
NIFSTD and NeuroLex: A Comprehensive Ontology Development Based on Multiple Biomedical Ontologies and Community Involvement
1. NIF: A COMPREHENSIVE ONTOLOGY FOR
NEUROSCIENCE & PRACTICAL GUIDE FOR
DATA-ONTOLOGY INTEGRATION
Maryann E. MARTONE, Fahim IMAM, Anita Bandrowski,
Stephen LARSON, Georgio ASCOLI, Gordon SHEPHERD, Jeffery S. GRETHE,
Amarnath GUPTA
Univ. of California, San Diego, CA; George Mason Univ., Fairfax, VA; Yale Univ.,
New Haven, CT
February 8, 2011
Funded in part by the NIH Neuroscience
Blueprint HHSN271200800035C via NIDA.
NEUROSCIENCE INFORMATION FRAMEWORK
NIFSTD Ontologies neuinfo.org 1
3. NIF today
• ~30M data records from 68 databases,
NIF registry (3600 software tools,
databases etc) + full text publication
search
– Focus of development now is on
integration of data with literature
– Better search of data (SKOS?)
• Annotation of data, now automated, will
become slightly more manual (we will assert
the contents of columns that match parts of
ontologies)
NIFSTD Ontologies neuinfo.org 3
4. NIF: DISCOVER AND UTILIZE WEB-BASED
NEUROSCIENCE RESOURCES
A portal to finding and
using neuroscience
resources
A consistent framework
for describing resources
Provides simultaneous
search of multiple types
of information, organized
by category
NIFSTD Ontology, a
critical component
Enables concept-based search
UCSD, Yale, Cal Tech, George Mason, Harvard MGH
Supported by NIH Blueprint
NIFSTD Ontologies neuinfo.org 4
5. NIF ‘dips’ into the lexicon for general
searches like ‘cerebellum’ or
‘ontology,’ where users can contribute
knowledge, but bring data into the
lexicon by using a application that
calls our web services
6. NIF today
• Ontology-based search
– Search requires all search terms: synonyms/acronyms/lexical variation
– Added gene: and other : searches are coming (toxin: drug:)
– Application logic: String match to multiple ontology terms = bring back
all (e.g., striatum and caudate putamen)
– Collapse duplicate classes by bridge files: same as relationship (Fahim)
– Heavy use of defined classes (GABAergic neuron, hippocampal neuron,
drug of abuse etc)
8. Cow example?
• Description of nose vs. tail:
which is more valid?
• Should they point to the same
entity?
• Is a mapping file the right place
to keep the knowledge that class
A is related to class B, or should
we assert sameness with
Mireot?
vs.
9. NIF STANDARD ONTOLOGIES (NIFSTD)
• Set of modular ontologies
– Covering neuroscience relevant
terminologies
– Comprehensive 50,000+ distinct
concepts + synonyms
• Expressed in OWL-DL language
• Closely follows OBO community
best practices
– As long as they seem practical
• Avoids duplication of efforts
– Standardized to the same upper level
ontologies, e.g.,
– Basic Formal Ontology (BFO), OBO
Relations Ontology (OBO-RO),
Phonotypical Qualities Ontology (PATO)
– Relies on existing community ontologies
e.g., CHEBI, GO, PRO, OBI etc.
9NIF Standard Ontologies
• Modules cover orthogonal domain
e.g. , Brain Regions, Cells, Molecules,
Subcellular parts, Diseases,
Nervous system functions, etc.
Bill Bug et al.
NIFSTD Ontologies neuinfo.org 9
10. ABOUT ONTOLOGY
• “Explicit specification of conceptualization”
- Tom Gruber
• Organizing the concepts involved in a domain
into a hierarchy and
• Precisely specifying how the concepts are
‘related’ with each other (i.e., logical axioms)
• Explicit knowledge are asserted but implicit
logical consequences can be inferred
– A powerful feature of an ontology
10NIF Standard Ontologies
NIFSTD Ontologies neuinfo.org 10
11. Class name Asserted necessary conditions
Cerebellum Purkinje cell 1. Is a ‘Neuron’
2. Its soma lies within 'Purkinje cell layer of cerebellar cortex’
3. It has ‘Projection neuron role’
4. It uses ‘GABA’ as a neurotransmitter
5. It has ‘Spiny dendrite quality’
Class name Asserted defining (necessary & sufficient) expression
Cerebellum neuron Is a ‘Neuron’ whose soma lies in any part of the ‘Cerebellum’ or
‘Cerebellar cortex’
Principal neuron Is a ‘Neuron’ which has ‘Projection neuron role’, i.e., a neuron
whose axon projects out of the brain region in which its soma lies
GABAergic neuron Is a ‘Neuron’ that uses ‘GABA’ as a neurotransmitter
ONTOLOGY – ASSERTED KNOWLEDGE
11NIF Standard Ontologies
NIFSTD Ontologies neuinfo.org 11
12. NIFSTD CURRENT VERSION
12NIF Standard Ontologies
• Key feature: Includes a set useful defined concepts
to have inferred classifications of asserted concepts
NIFSTD Ontologies neuinfo.org 12
14. CONCEPT-BASED SEARCH
• Search Google: GABAergic neuron
• Search NIF: GABAergic neuron
– NIF automatically searches for types of
GABAergic neurons
Types of GABAergic
neurons
NIFSTD Ontologies neuinfo.org 14
15. NIFSTD AND NEUROLEX WIKI
• Semantic wiki platform
• Provides simple forms for
structured knowledge
• Can add concepts,
properties
• Generate hierarchies
without having to learn
complicated ontology tools
• Good teaching tool for
principles behind
ontologies
• Community can contribute
– Each term gets its own unique
ID
NIF Standard Ontologies
15
Stephen D. Larson et al.
NIFSTD Ontologies neuinfo.org 15
16. ACCESS TO SHARED ONTOLOGIES
• NIFSTD is available as
– OWL Format http://ontology.neuinfo.org
– RDF and SPARQL Endpoint
• Specific contents through web services
– http://ontology.neuinfo.org/ontoquest.html
• Available through NCBO Bioportal
– Repository of biomedical ontologies
– 199 ontologies including NIFSTD
– Provides annotation and mapping services
– http://bioportal.bioontology.org/
• INCF Program on Ontologies for
Neural Structure
– Neuronal Registry Task Force: Description of
neural properties
– Structural Lexicon: Description of structures
across scales
NIF Standard Ontologies 16
NIFSTD Ontologies neuinfo.org 16
17. NIF Standard Ontologies 17
Domain External Source Import/
Adapt
NIFSTD
Module
Organism
taxonomy
NCBI Taxonomy, GBIF, ITIS, IMSR, Jackson Labs mouse catalog; . Specifically
the taxonomy of model organisms in common use by neuroscientists
Adapt NIF-Organism
Molecules IUPHAR ion channels and receptors, Sequence Ontology (SO); pending: NCBI,
NCBI Entrez Protein, NCBI RefSeq, NCBI Homologene; NIDA drug lists, ChEBI,
and Protein Ontology (PRO)
Adapt
IUPHAR;
import PRO
NIF-Molecule
NIF-Chemical
Sub-cellular Sub-cellular AnatomyOntology (SAO). Extracted cell parts and subcellular
structures from SAO-CORE . Soon to be importing GO Cellular Component with
mapping
Import NIF-Subcellular
Cell CCDB, NeuronDB, NeuroMorpho.org . terminologies; pending: OBO Cell
Ontology
Adapt NIF-Cell
Gross Anatomy NeuroNames extended by including terms from BIRN, SumsDB, BrainMap.org,
etc;
Multi-scale representation of Nervous System Mac Macroscopic anatomy
Adapt NIF-
GrossAnatomy
Nervous system
function
Sensory, Behavior, Cognition terms from NIF, BIRN, BrainMap.org, MeSH, and
UMLS
Adapt NIF-Function
Nervous system
dysfunction
Nervous system disease from MeSH, NINDS terminology; pending: OMIM Adapt/Import NIF-
Dysfunction
Phenotypic
qualities
PATO Imported as part of the OBO foundry core Import NIF-Quality
Investigation:
reagents
Overlaps with molecules above, especially RefSeq for mRNA, ChEBI, Sequence
ontology; pending: Protein Ontology
import NIF-
Investigation
Investigation:
instruments,
protocols, plans
Based on Ontology for Biomedical Investigation (OBI ) to include entities for
biomaterial transformations, assays, data collection, data transformations.
Adapt NIF-
Investigation
Investigation:
resource type
NIF, OBI, NITRC, Biomedical Resource Ontology (BRO) Adapt NIF-Resource
Biological
Process
Gene Ontology’s (GO) biological process in whole Import NIF-
BioProcess
NIFSTD EXTERNAL COMMUNITY SOURCES
NIFSTD Ontologies neuinfo.org 17
18. • So Far..
– Overlaps are detected and mappings were carefully
curated
– Included a bridging module that asserts equivalencies
between NIF-Dysfunction and DOID
• We could MIREOT DOID Classes as well
• Drawback was loosing NIF’s annotation properties.
• Having the bridgeing module allowed us to have contents
from both ontologies and to keep the mappings as well. (Did
the same with NIF-Subcellular and GO-Cell Component)
• Collaborating on Mental Disorder - Addiction/
Substance related disorder with DOID group
• Taking a look at Barry Smith’s paper on Foundations for a
realist ontology of mental disease
NIF Standard Ontologies 18
NIFSTD AND DOID COLLABORATION
20. SUMMARY AND CONCLUSIONS
• NIF project with NIFSTD is an example of how
ontologies can be used to enhance search and
data integration across diverse resources
• NIFSTD continues to create an increasingly rich
knowledgebase for neuroscience integrating with
other life science community
• NIF encourages the use of community ontologies
for resource providers
20NIF Standard Ontologies
NIFSTD Ontologies neuinfo.org 20
21. Some questions:
• If someone asserts sameness should that be
treated differently by others?
– How would we know? Should there be a tool that
would search these assertions?
• Can a lexicon be used as a set of base classes for
use in ontology building?
– We took this approach with nervous system cells by
adding properties, then asserted hierarchies:
• GABAergic neuron
• Cerebellum neuron
• Intrinsic neuron
NIFSTD Ontologies neuinfo.org 21
22. Even more questions?
• If a term has no definition, then should it exist
in the lexicon?
• Do tests belong in ontologies?
NIFSTD Ontologies neuinfo.org 22
23. • NIFSTD Ontologies
http://ontology.neuinfo.org
• NeuroLex Wiki
http://neurolex.org
• Neuroscience Information Framework
(NIF)
http://neuinfo.org
23NIF Standard Ontologies
NIFSTD Ontologies neuinfo.org 23
Editor's Notes
A critical component of Neuroscience Information Framework project (http://neuinfo.org), NIF Standard (NIFSTD) is a set of modular ontologies covering a comprehensive set of neuroscience terminologies. This highlight the key features of NIFSTD and how NIF uses it to enable an effective concept-based search against a diverse collection of neuroscience resources over the web. Closely follows the best practices of Open Biological Ontology (OBO) community, And is standardized to the same upper level ontologies for biomedical sciences and promotes easy extensionNIFSTD is designed to collate existing neuroscience terminologies into a coherent set of orthogonal and interoperable modules. Relies on existing ontologies as the initial building block e.g., CHEBI, GO, PRO, OBI etc.
Within NIF-Cell module, along with other neurons, we have cerebellum neuron and we say that it’s a neuron. We also assert other things that are known about this type of neuron.
We have recently released NIFSTD v.1.8 (http://ontology.neuinfo.org) where the key feature is the inclusion of various cross-domain bridge modules. These modules contain necessary restrictions along with a set of defined classes to infer useful classification of neurons and molecules. These classifications include neurons in terms of their soma locations in different brain regions (e.g., Hippocampal neurons, Cerebellum neurons), neurons by their neurotransmitter (e.g., GABAergic neuron) and circuit roles (e.g., intrinsic neurons), classification of molecules and chemicals by their molecular roles (e.g., Drug of abuse, Neurotransmitter). We keep the logical restrictions and definitions on required set of classes assigned in a separate bridge file so that the core universal hierarchies in different modules like NIF-Cell, NIF-Molecule, or NIF-Anatomy are open to easy extensions for broader communities without worries about specific, NIF-centric views. It keep the modularity principles intact and useful. You can always exclude the bridging module just to focus on the core module and build your own restrictions that you think is appropriate for your application.
Having the defined classes enabled us to have useful concept-based queries through the NIF search interface. For example, while searching for ‘GABAergic neuron’, the system recognizes the term as ‘defined’ from the ontology, and looks for any neuron that has GABA as a neurotransmitter (instead of the lexical match of the search term) and enhances the query over those inferred list of neurons.
One of the largest roadblocks that we encountered was the lack of tools for the neuroscience community to contribute their knowledge into a formal ontology like NIFSTD. NIF has created NeuroLex (http://neurolex.org), a semantic wiki interface for the domain experts as an easy entry point to the NIFSTD contents. It has been extensively used in the area of neuronal cell types where NIF is working with a group of neuroscientists to create a comprehensive list of neurons and their properties. While the properties in NeuroLex are meant for easier interpretation, the restrictions in NIFSTD are more rigorous and based on standard OBO-RO relations.We organize our known knowledge within NeuroLex i.e., what can we say about a concept e.g., Cerebellum Purkeje neuron is a neuron whose soma resides within Cerebellum brain region, it has_role projection neuron role, it has GABA as a Neurotransmitter and so on.
Ontologies available as OWL file, RDF and through Web Serviceshttps://confluence.crbs.ucsd.edu/display/NIF/OntoQuestMain.NCBO Bioportal (http://bioportal.bioontology.org/)Repository of ontologies for biomedical research199 ontologies (including NIFSTD)Contains many mappingsProvides annotation servicesINCF Program on Ontologies for Neural StructuresNeuronal Registry Task ForceDescription of neural propertiesStructural LexiconDescription of properties across scales
To initiate the process of finding the overlaps between NIFSTD and DO, I have extracted a list of possible terms along with their IDs into the attached spreadsheet. The extraction process was automatic and based on maximum lexical similarity between the term lebels or synonymous terms (based on BioPortal mapping script). There are some obviouse non-sensical mapping that needs to be excluded from the list. Here are few things to consider:1. The term 'Cancer' refers to an Organism in NIFSTD and therefore should not be mapped with DO's 'Cancer'.2. Some of the terms in DO matches with GO, PATO, SO, and NIF-Anatomy. We need to decide on those as to where should they belong natively, or even if those mappings are even valid. Some of the term labels in DO requires some revision e.g., 'paralysed' is a PATO quality; while it's synonym from DO did match with NIFSTD's 'Paralysis', I'm not sure why 'paralyzed' should be the label.3. Some of the terms from NIFSTD maps with multiple terms with DO due to their synonymous similariuty. We need to go over them and decide the distinctions. For example, NIFSTD's 'Depressive Disorder' maps with all four of the following terms from DO: DOID: endogenous depression DOID: melancholia DOID: neurotic depression DOID: unipolar depressionSeems like the synonyms for the term in NIFSTD are not valid and therefore be corrected.
The NIF project provides an example of how ontologies can be used to enhance search and data integration across diverse resources. As the project moves forward, we are using NIFSTD to build an increasingly rich knowledgebase for neuroscience that integrates with the larger life science community