an ontology-based approach for computational phenomics: application to autism spectrum disorder

Stanford Centerfor Biomedical Informatics Research

An Ontology-Based Approach for Computational Phenomics:

Application to Autism Spectrum Disorder

Amar K. Das, MD, PhDDepartments of Medicine and

of Psychiatry and Behavioral Sciences

NCBO WebinarOctober 7, 2009

Outline Motivations NDAR project Phenologue project Future Directions


Motivation

Psychiatric Genetics

Phenotyping

TerminologyOntology

Logic


Hasler G,et al. Toward constructing an endophenotype strategy for bipolar disorders. Biological Psychiatry (2006)

Represent findings and their links using structured knowledge


Phenomics

“A primary task for the new field of phenomics will be to clarify what, in practical terms, constitutes a phenotype and then to delineate the different phenotypic components that compose the phenome.”

Freimer & Sabatti, Nature Genetics (2003)


OMIM


dbGaP

Mailman, M.D. Nature Genetics (2007)


PhenoWiki


Current Approaches Lack of standardization Lack of organization Lack of computability


Autism DSM-IV DiagnosisA total of six (or more) items from (1), (2), and (3), with at least two from (1), and one each from (2) and (3)

(1) qualitative impairment in social interaction, as manifested by at least two of the following:a) marked impairments in the use of multiple nonverbal behaviors such as eye-to-eye gaze, facial expression, body posture, and gestures to regulate social interactionb) failure to develop peer relationships appropriate to developmental levelc) a lack of spontaneous seeking to share enjoyment, interests, or achievements with other people, (e.g., by a lack of showing, bringing, or pointing out objects of interest to other people)d) lack of social or emotional reciprocity


Autism DSM-IV Diagnosis(2) qualitative impairments in communication as manifested by at least one of the following:a) delay in, or total lack of, the development of spoken language (not accompanied by an attempt to compensate through alternative modes of communication such as gesture or mime)b) in individuals with adequate speech, marked impairment in the ability to initiate or sustain a conversation with othersc) stereotyped and repetitive use of language or idiosyncratic languaged) lack of varied, spontaneous make-believe play or social imitative play appropriate to developmental level


Autism DSM-IV Diagnosis(3) restricted repetitive and stereotyped patterns of behavior,interests and activities, as manifested by at least two of the following:a) encompassing preoccupation with one or more stereotyped and restricted patterns of interest that is abnormal either in intensity or focusb) apparently inflexible adherence to specific, nonfunctional routines or ritualsc) stereotyped and repetitive motor mannerisms (e.g hand or finger flapping or twisting, or complex whole body movements)d) persistent preoccupation with parts of objects

Delays or abnormal functioning in at least one of the following areas, with onset prior to age 3 years:(1) social interaction(2) language as used


NDAR (ndar.nih.gov)


Goals of NDAR Develop standards to promote meta-

analyses and cross site research data comparisons

Provide researchers access to useful software tools and infrastructure

Promote the sharing of research data relevant to ASD


NIH Research Support in Autism $100 million/year in funding

Investigator-initiated grants (R01’s) Special initiatives, e.g. RFA for genetics Centers and networks Training grants (To institutions and individuals)

New initiatives Intramural Research Program on Autism Autism Centers of Excellence (ACE) National Database for Autism Research (NDAR) ARRA stimulus program


BIRN Mediator


Query and Reporting

BIRN Services& Resources

NDAR System

Security

Portal

Grid Computing

Collaboration

Data Storage Management

Data Integration Tools

AuditingUser Management

Subject Tracking & Management

Clinical Assessments(OpenClinica)

Common Measures

Study Management

Neuroimaging

Image Analysis

Image Processing

Image data access

Genomics

Genomics data access

Data Integration


NDAR Codebook


Phenotypes in Psychiatry

‘The observable structural and functional characteristics of an organism determined by its genotype and modulated by its environment’

Diagnostic component Intermediate phenotype Quantitative phenotype Covariates


Example Query #1

Find all subject who are verbal (ADIR A14). Then look at their IQ (Cognitive Total IQ > 70) and whether or not they have seizures (Medical History Q10). Also find out if they have an abnormal MRI or any genetic abnormalities.


Example Query #2

Use head circumference to categorize macroencephaly. Then see if the subjects differ in their ADOS, ADI-R, cognitive, and language profiles, and combine this with genetic data.


NDAR Project Systematic Review Ontology Development Database Infrastructure


Systematic Review “(ADI-R or ADOS or Vineland) and

(genes or genetics) and autism” 26/43 papers relevant Mean # phenotypes 4.1, range 1-13 Three basic types (1:1, sum, cutoff score)

Tu, S. W. AMIA Annual Proceedings (2008)


Systematic Review Different terms

e.g., ‘age of first phrases’ and ‘age of onset of phrase speech’

Different cutoff scorese.g., ‘delayed word’

Different definitionse.g., ‘regression’e.g., use of different instruments


Clinical Research Study

Clinical Trial StudyCase Study

Controlled Case Study Study Arms

Ontology A taxonomy with multiple link types,

each with precise meaning


Perspectives on ‘Ontology’ Philosophy: The study

of what entities and what types of entities exist in reality

Computer Science: A schema that represents a domain and is used to reason about the objects in that domain and the relations between them


Critical to the ‘Semantic Web’ Shared research and development plan to

Provide explicit semantic meaning to data and knowledge shared on the Web

Bring structure to Web content Advance the current state-of-the-art in Web

information retrieval, which is keyword searching

Distributed applications will be able to process data and knowledge automatically through the use of ontologies


OWL: Web Ontology Language Advances current Semantic Web standards

by using ontologies to represent knowledge OWL can be used to build ontologies of

high-level descriptions, based on three concepts: Classes (e.g., Subject, Phenotype, Genotype) Properties (e.g., isBearerOf, hasResults) Individuals (e.g., “Macroencephaly”)


SubjectGenotype

Phenotype

mutIn-RELN

Macro-encephaly

011451

hasResult

isBearerOf

OWL: Web Ontology Language


BIRNLex A controlled terminology for annotation of

BIRN data sources, focusing on imaging data from human subjects and mouse models

Terms cover neuroanatomy, molecular species, behavioral and cognitive processes, subject information, experimental practice and design


Basic Formal Ontology An upper ontology which can be used

to support the development of domain ontologies used in scientific research

All concepts are subclasses of Continuants: exists in full at any time in

which it exists at all Occurants: has temporal parts and that

happens, unfolds or develops through time


OBO Foundry Ontologies should be orthogonal

Minimize overlap Each distinct entity type (universal) should

only be represented once Partition efforts in the OBO Foundry

rationally to help organize and coordinate the ontology development


CONTINUANT OCCURRENT RELATION TO

TIME GRANULARITY INDEPENDENT DEPENDENT

ORGAN AND ORGANISM

Organism (NCBI

Taxonomy)

Anatomical Entity (FMA, CARO)

Organ Function (FMP, CPRO)

Organism-Level Process

(GO)

CELL AND CELLULAR

COMPONENT

Cell (CL)

Cellular Component (FMA,GO)

Cellular Function

(GO)

Phenotypic Quality (PaTO)

Cellular Process (GO)

MOLECULE Molecule

(ChEBI, SO, RnaO, PrO)

Molecular Function (GO)

Molecular Process (GO)

Chris Mungall, PATO


SWRL: Semantic Web Rule Language W3C specification for expressing

logical rules that can be formulated in terms of OWL concepts

Rules in SWRL can be used to deduce new knowledge about an existing OWL ontology

Specification can be extended through the use of built ins


hasParent(?x, ?y) ^ hasBrother(?y, ?z)→ hasUncle(?x, ?z)

Example SWRL Rule: hasUncle


Example SWRL Rule: hasSister

Person(Amar) ^ hasSibling(Amar, ?s)

^ Woman(?s)

→ hasSister(Amar, ?s)


Person(?p) ^ hasAge(?p,?age) ^ swrlb:lessThan(?age,17) → Child(?p)

Example SWRL Rule: Child


Rule-Based Methods Extensions to SWRL

Temporal Library of temporal built ins

Query Extraction of results as a table

MakeSet Support for set-based operations


Development Methods Extensions to BIRNLex Encoding of phenotypes Querying of NDAR database


Autism Assessment Result

Figure 1. The representation of data collected through the ADI-2003 autism assessment instrument as part of the autism ontology.


Phenotype Representation

Figure 2. The representation of the Status of age of words phentotype group as a OWL class partition by the possible statuses.


Phenotype Rule

ADI_2003_result(?assessment) ^

acqorlossoflang_aword(?assessment,?wordage) ^

swrlb:greaterThan(?wordage, 24) ^

subject_id(?assessment, ?subjectId) ^

orgtax:Human(?subject) ^

subject_id(?subject, ?subjectId)

→ birn_obo_ubo:bearer_of(?subject, Delayed_word)


Phenotype Rules


Ontology-Driven Querying

Young, L. IEEE CBMS (2009)


Phenologue Project Develop an ontology of endophenotypes that maps brain

connectivity, neural deficits, and genetic markers into a subject domain theory

Develop logic-based methods to encode and classify endophenotypes based on multi-scale measurements

Create tools to acquire new endophenotypes and annotate phenotype-genotype findings in online resources such as published literature

Develop query-elicitation methods that can evaluate hypotheses about the subject domain theory of endophenotypes using deductive inference


Phenologue Project

Database

Phenotype Definitions

New Associations

Query

Catalog Analysis


Rule Technologies Rule paraphrasing Rule elicitation Rulebase visualization Knowledge mining using rules


Rule Paraphrasing


Rule Elicitation


Rulebase Visualization


Computational Phenomics Informatics methods to support

phenomics Apply machine learning methods to

discover groups of rules with common semantics

Use natural language processing method to discover phenotype rules in published text


Semantic Similarity


Future Directions Expand phenotype categories Use natural language processing

method to discover phenotype rules in published text

Apply machine learning methods to discover groups of rules with common semantics


Summary The development of a standardized,

organized, and computable set of phenotype terms is central to etiologic studies of complex disorders

The use of ontologies and rules to model phenotypes is feasible and can enable automated discovery of new phenotype-genotype relationships


Acknowledgments Stanford Group

Martin O’Connor Saeed Hassanpour Duriel Hardy Ravi Shankar Lakshika Tennakoon Samson Tu

National Center for Biomedical Ontology Mark Musen Daniel Rubin

NDAR/NIMH Lynn Young Matthew McAuliffe Dan Hall Lisa Gilotty

Biomedical Informatics Research Network Bill Bug Maryann Martone

an ontology-based approach for computational phenomics: application to autism spectrum disorder

Documents