You are on page 1of 39

A

Seman(c Web for Neuroscience? What could that mean?


Alan Ru(enberg Science Commons h(p://neurocommons.org/

Alan Ruttenberg!

Neuroinformatics 2009!

Science Commons
Science Commons is a project of Crea(ve Commons
Crea-ve Commons provides free tools that let authors, scien-sts, ar-sts, and educators easily mark their crea-ve work with the freedoms they want it to carry 140,000,000 objects on the Web under CC licenses in 40+ countries 800+ peer-reviewed journals carry CC licensing, including Public Library of Science For consumers of knowledge: make it easy to use and re-use informa-on and increase chances for discovery For providers of knowledge: provide legal certainty and automated aHribu-on and tracking For funders: improve return on investment by encouraging and enabling re- use
Neuroinformatics 2009!

Science Commons specializes CC to science


Alan Ruttenberg!

Goals of my talk
Give you a sense for what is mean by Seman-c Web Show glimpses of what we are aiming for Introduce the why, what and how of Ontology Talk about the future
Where we are heading How you can par-cipate.
Alan Ruttenberg!
Neuroinformatics 2009!

Cut and paste on the Seman-c Web for Science SPARQL for specic answers to precise ques-ons

The Seman-c Web in a Nutshell


Adds to Web standards and prac(ces encouraging
Unambiguous names for things, classes, and rela-onships Well organized and documented in ontologies With data expressed using uniform knowledge representa-on languages To enable computa-onally assisted exploita-on of informa-on That can be easily integrated from dierent sources Both within and across public and organiza-onal boundaries

Alan Ruttenberg!

Neuroinformatics 2009!

Why should you want a Seman-c Web?


Provides opportuni-es for clearer communica-on of research results Encourages separa-on of concerns and par-cipa-on in research by experts in each aspect of mul-faceted work Encourages network eects surprises of combining resources, invi-ng experiment Presents opportunity for eciencies at global scales.
Alan Ruttenberg!
Neuroinformatics 2009!

Copy/Paste on the Semantic Web for Science


http://sparql.neurocommons.org/map/#Kcnip3@2850,Kcnd1@2800!Allen

Javascript

Brain Institute Servers

http://www.brainmap.org://.0205032816_B.aff/TileGroup3/1-0-1.jpg

SPARQL AJAX

Query

Neurocommons Servers
Alan Ruttenberg!
Neuroinformatics 2009!

URL
Google Maps API

BIRN: viewed source, use our code in their own applica-ons

Alan Ruttenberg!

Neuroinformatics 2009!

Specic answers to precise ques-ons


Signal transduc-on pathways are considered to be rich in druggable targets - proteins that might respond to chemical therapy CA1 Pyramidal Neurons are known to be par-cularly damaged in Alzheimers disease. Cas-ng a wide net, can we nd candidate genes known to be involved in signal transduc-on and ac-ve in Pyramidal Neurons?

Alan Ruttenberg!

Neuroinformatics 2009!

Google: 223,000 results

Alan Ruttenberg!

Neuroinformatics 2009!

Pubmed: 2800 results

Alan Ruttenberg!

Neuroinformatics 2009!

A SPARQL query for processes involved in pyramidal neurons


prex go: <hHp://purl.org/obo/owl/GO#> prex rdfs: <hHp://www.w3.org/2000/01/rdf-schema#> prex owl: <hHp://www.w3.org/2002/07/owl#> prex mesh: <hHp://purl.org/commons/record/mesh/> prex sc: <hHp://purl.org/science/owl/sciencecommons/> prex ro: <hHp://www.obofoundry.org/ro/ro.owl#> select ?gene_name ?process_name where { ?pubmed_record ?p mesh:D017966 . ?ar-cle sc:iden-ed_by_pmid ?pubmed_record. ?gene_record sc:describes_gene_or_gene_product_men-oned_by ?ar-cle. ?protein rdfs:subClassOf ?restric-on. ?restric-on owl:onProperty ro:has_func-on. ?restric-on owl:someValuesFrom ?restric-on2. ?restric-on2 owl:onProperty ro:realized_as. ?restric-on2 owl:someValuesFrom ?process. {{?process ro:part_of go:GO_0007166} union {?process rdfs:subClassOf go:GO_0007166 }} ?protein rdfs:subClassOf ?parent. ?parent owl:equivalentClass ?restric-on3. ?restric-on3 owl:onProperty sc:is_protein_gene_product_of_dna_described_by. ?restric-on3 owl:hasValue ?gene_record. ?gene_record rdfs:label ?gene_name. ?process rdfs:label ?processname. }

Mesh: Pyramidal Neurons

Pubmed: Journal Articles

Entrez Gene: Genes

GO: Signal Transduction

Inference required

Alan Ruttenberg!

Neuroinformatics 2009!

Results
DRD1, 1812 ADRB2, 154 ADRB2, 154 DRD1IP, 50632 DRD1, 1812 DRD2, 1813 GRM7, 2917 GNG3, 2785 GNG12, 55970 DRD2, 1813 ADRB2, 154 CALM3, 808 HTR2A, 3356 DRD1, 1812 SSTR5, 6755 MTNR1A, 4543 CNR2, 1269 HTR6, 3362 GRIK2, 2898 GRIN1, 2902 GRIN2A, 2903 GRIN2B, 2904 ADAM10, 102 GRM7, 2917 LRP1, 4035 ADAM10, 102 ASCL1, 429 HTR2A, 3356 ADRB2, 154 PTPRG, 5793 EPHA4, 2043 NRTN, 4902 CTNND1, 1500 adenylate cyclase ac-va-on adenylate cyclase ac-va-on arres-n mediated desensi-za-on of G-protein coupled receptor protein signaling pathway dopamine receptor signaling pathway dopamine receptor, adenylate cyclase ac-va-ng pathway dopamine receptor, adenylate cyclase inhibi-ng pathway G-protein coupled receptor protein signaling pathway G-protein coupled receptor protein signaling pathway G-protein coupled receptor protein signaling pathway G-protein coupled receptor protein signaling pathway G-protein coupled receptor protein signaling pathway G-protein coupled receptor protein signaling pathway G-protein coupled receptor protein signaling pathway G-protein signaling, coupled to cyclic nucleo-de second messenger G-protein signaling, coupled to cyclic nucleo-de second messenger G-protein signaling, coupled to cyclic nucleo-de second messenger G-protein signaling, coupled to cyclic nucleo-de second messenger G-protein signaling, coupled to cyclic nucleo-de second messenger glutamate signaling pathway glutamate signaling pathway glutamate signaling pathway Many of the genes are indeed related glutamate signaling pathway integrin-mediated signaling pathway Alzheimers Disease through gamma nega-ve regula-on of adenylate cyclase ac-vity secretase (presenilin) activity! nega-ve regula-on of Wnt receptor signaling pathway Notch receptor processing Notch signaling pathway serotonin receptor signaling pathway transmembrane receptor protein tyrosine kinase ac-va-on (dimeriza-on) transmembrane receptor protein tyrosine kinase signaling pathway transmembrane receptor protein tyrosine kinase signaling pathway transmembrane receptor protein tyrosine kinase signaling pathway Wnt receptor signaling pathway

to

Alan Ruttenberg!

Neuroinformatics 2009!

But answers are also available by a HTTP GET


/sparql/?query=PREFIX%20owl%3A%20%3Chttp%3A%2F%2Fwww.w3.org %2F2002%2F07%2Fowl%23%3E%0APREFIX%20go%3A%20%3Chttp%3A%2F %2Fpurl.org%2Fobo%2Fowl%2FGO%23%3E%0APREFIX%20obo%3A %20%3Chttp%3A%2F%2Fwww.geneontology.org%2Fformats%2FoboInOwl %23%3E%0APREFIX%20rdfs%3A%20%3Chttp%3A%2F%2Fwww.w3.org %2F2000%2F01%2Frdf-schema%23%3E%0A%0Aselect%20%20%3Fname %20%20%3Fclass%20%3Fdefinition%0Afrom%20%3Chttp%3A%2F %2Fpurl.org%2Fcommons%2Fhcls%2F20070416%3E%0Awhere%0A%7B %20%20%20graph%20%3Chttp%3A%2F%2Fpurl.org%2Fcommons%2Fhcls %2F20070416%2Fclassrelations%3E%0A%20%20%20%20%20%7B%3Fclass %20rdfs%3AsubClassOf%20go%3AGO_0008150%7D%0A %20%20%20%20%3Fclass%20rdfs%3Alabel%20%3Fname.%0A %20%20%20%20%3Fclass%20obo%3AhasDefinition%20%3Fdef.%0A %20%20%20%20%3Fdef%20rdfs%3Alabel%20%3Fdefinition%20%0A %20%20%20%20filter(regex(%3Fname%2C%22%5BDd%5Dendrite%22))%0A %7D%0A!

So someone else can build something beHer


Alan Ruttenberg!
Neuroinformatics 2009!

Use interface toolkits e.g. Exhibit are being built to accept RDF/SPARQL

Credit: Eric Neumann Alan Ruttenberg!


Neuroinformatics 2009!

For what neurological disorders are cell lines available? For Parkinsons disease, what -ssue and cell lines are available? Give me informa-on on the receptors and channels expressed in cor-cal neurons What chemical agents can be used visualizing the nervous system?

We are always asking ques-ons

A ques-on I was asked


Create a system that will let us prioritize an expected 2000 siRNA hits according to whether there is chemical matter for studying them, e.g. validated antibodies, since we can only follow up on 600. We know how to use Semantic Web technology to answer these kinds of questions (but there is no free lunch) Alan Ruttenberg!
Neuroinformatics 2009!

A moment in the life of an ontologist: Untangling terms that mix up func-on and material
Ligand NeurotransmiHer Hormone Pep-de

Looking for pep<des?

Alan Ruttenberg!

Neuroinformatics 2009!

Refactoring to maximize connec-ons


Pep<deReceptorLigand - A pep(de that has a func-on which makes it able to bind to a receptor Pep<deNeurotransmi(er - A pep(de expressed in a neuron that has a func-on which makes it able to regulate another neuron Pep<deHormone - A pep(de that produced in one organ and having an regulatory eect in another organ. Pep$de - A short polymer of amino acids

Looking for pep<des?


Neuroinformatics 2009!

Alan Ruttenberg!

Pep-des from CHEBI Chemical En--es of Biological Interest

Alan Ruttenberg!

Neuroinformatics 2009!

Hormone Ac<vity from GO Molecular Func-on

Alan Ruttenberg!

Neuroinformatics 2009!

Alan Ruttenberg!

Neuroinformatics 2009!

From English to OWL


ALL instances of Pep$deHormone are an instance of Pep$de that has_role SOME instance of HormoneAc$vity

Alan Ruttenberg!

Neuroinformatics 2009!

From English to OWL

chebi:25905 = <hGp://purl.org/obo/owl/CHEBI#CHEBI_25905>
Alan Ruttenberg!
Neuroinformatics 2009!

Each URI provides a way to access useful informa-on for people and machines

Alan Ruttenberg!

Neuroinformatics 2009!

Whats really at hHp://purl.org/obo/OBI_0000225 Dont read this (but machines love it!)
<?xml version="1.0" encoding="UTF-8"?> <?xml-stylesheet type="text/xsl" href="http://ashby.csail.mit.edu/cgi-bin/obiterm.xsl?ref=OBI_0000225"?> <owl:Class rdf:about="&obo;OBI_0000225"> <obo:OBI_0000274 rdf:datatype="&xsd;string">Jennifer Fostel</obo:OBI_0000274> <obo:OBI_0000275 rdf:datatype="&xsd;string">site is a material (building) having the role site</obo:OBI_0000275> <obo:OBI_0000275 rdf:datatype="&xsd;string">solution1: is a physical location, should maybe go under processual context, and then be used in conjunction with the located relation</obo:OBI_0000275> <obo:OBI_0000275 rdf:datatype="&xsd;string">solution2: site is related to trial used located_in relation site can bear the role</obo:OBI_0000275> <obo:OBI_0000281 rdf:resource="&obo;OBI_0000320"/> <obo:OBI_0000287 rdf:datatype="&xsd;string">A field, a laboratory, a medical institute, a pharmaceutical company</ obo:OBI_0000287> <obo:OBI_0000288 rdf:datatype="&xsd;string">Investigation site</obo:OBI_0000288> <obo:OBI_0000291 rdf:datatype="&xsd;string">Investigation site is a role born by a site or material realized in an investigation which is located at the investigation site</obo:OBI_0000291> <dc:creator rdf:resource="&obo;obi/project"/> <rdfs:isDefinedBy rdf:resource="&obo;obi.owl"/> <rdfs:label rdf:datatype="&xsd;string">investigation site role</rdfs:label> <rdfs:subClassOf rdf:resource="http://www.ifomis.org/bfo/1.1/snap#Role"/> </owl:Class><doap:Project rdf:about="&obo;obi/project"> <doap:browse rdf:resource="http://obi.svn.sourceforge.net/viewvc/obi/"/> <doap:bug-database rdf:resource="http://sourceforge.net/tracker/?group_id=177891&amp;atid=886178"/> <doap:homepage rdf:resource="http://obi.sourceforge.net/"/> <doap:mailing-list rdf:resource="mailto:obi-devel@groups.google.com"/> <doap:release rdf:resource="&obo;obi/version-2008-03-10"/> <doap:repository rdf:resource="http://obi.svn.sourceforge.net/svnroot/obi/"/> <doap:wiki rdf:resource="https://wiki.cbil.upenn.edu/obiwiki/index.php?title=HomePage"/> </doap:Project> <doap:Version rdf:about="&obo;obi/version-2008-03-10"> <doap:file-release rdf:resource="http://purl.org/obo/2008-03-10/obi.owl"/> <doap:file-release rdf:resource="http://purl.org/obo/obi.owl"/> </doap:Version> </rdf:RDF> Alan Ruttenberg!

Neuroinformatics 2009!

NeuronDB
Started with homegrown ontology. Problem: How to link with anything else Eg. No links to evidence, receptors versus proteins with receptor ac-vity (like GOA) Process: xing OWL, understanding GO and making representa-on compa-ble Augmented what was in ontology, e.g. provenance Ended with something that linked with GO Func-on and, via that, the rest of the Neurocommons Takeaway: Having your own ontology: Nice. Being able to connect your data to the rest of knowledge: Priceless!
Alan Ruttenberg!
Neuroinformatics 2009!

Inconsistency!
Once NeuronDB was represented using this methodology, and an OWL reasoner was run, it declared the ontology inconsistent Problem: There are contradictory asser-ons about whether a par-cular ionic current occurs in a par-cular cell type. What to do? Inconsistency is NOT acceptable, but this can be resolved by thinking about a bit about what the resources is trying to accomplish.
Alan Ruttenberg!
Neuroinformatics 2009!

Three Ways of Represen-ng Scien-c Knowledge


Record level: Represent database records. Inconsistent if two sources disagree about contents of a eld. Statement level: Represent what researchers say. Inconsistent if two people disagree about what a paper said Domain level: OBO Foundry approach. Represent your best understanding of our consensus of what exists. Inconsistent if facts contradict. We need all three (but make clear which is which)
Alan Ruttenberg!
Neuroinformatics 2009!

How to Build Consensus around Ontology


Understand what terms mean by rela-ng them (tracing them) to elements in reality. Or say when you are not. Know what the instances are Ask yourself how instances are related to each other Dene classes as groups of instances that have shared proper-es Document and organize this knowledge in a way that can be managed in a distributed manner The product of this eort is your Ontology

Alan Ruttenberg!

Neuroinformatics 2009!

OBO Foundry - What


A subset of OBO ontologies whose developers have agreed in advance to accept a common set of principles reec-ng best prac-ce in ontology development designed to ensure
-ght connec-on to the biomedical basic sciences compa-bility interoperability, common rela-ons formal robustness support for logic-based reasoning

Alan Ruttenberg!

Neuroinformatics 2009!

OBO Foundry - How


Collabora-on Peer review Design principles evolve as we learn AHribu-on at all levels Due credit Metrics to measure progress and eec-veness

Alan Ruttenberg!

Neuroinformatics 2009!

Selected OBO Ontologies


Gene Ontology (GO) Rela-on Ontology (RO) Chemical En--es of Biological Interest (CheBI) Phenotype Ontology (PATO) Ontology for Biomedical Inves-ga-ons (OBI) Cell Ontology (CL) Protein Ontology (PRO) Founda-onal Model of Anatomy (FMA) Sequence Ontology (SO) Subcellular Anatomy (SAO) Informa-on Ar-fact (IA0)
Neuroinformatics 2009!

Alan Ruttenberg!

Ini-al goals: Common set of iden<ers, terms, and deni<ons, iden-fying elements of neuroanatomic structure and types of neurons Three task forces:
Structural Lexicon Neuronal Registry Representa-on and Deployment

INCF Program for Ontology of Neural Structures (PONS)

Also liason with Digital Atlasing task force


Alan Ruttenberg!
Neuroinformatics 2009!

The NeuroCommons
Built on open resources: public domain, open databases, open literature Encoded using open architectures and technical standards OWL, SPARQL, OBO Promo-ng legal predictability and certainty
Licenses that are easy to use and understand That impose the lowest possible transac-on costs on users The support the freedom to integrate

Scou-ng out issues to be important in building a seman-c web for neuroscience

Alan Ruttenberg!

Neuroinformatics 2009!

Elements of the web of Neuroscience Very connected!


structure, connec-vity physiology, pathophysiology func-on, malfunc-on molecular mechanism disease, injury experimental materials and method pa-ents, clinical diagnosis, devices, treatments studies, measurements, images publica-ons models, simula-ons, hypotheses, predic-ons
Neuroinformatics 2009!

Alan Ruttenberg!

Challenges on the way to the Seman-c Web for Neuroscience


Neuroscience large domain, rapidly advance Building quality, durable, maintainable ontologies is dicult another developing science Sociology there are barriers to sharing we need to understand them and learn ways to overcome them

Alan Ruttenberg!

Neuroinformatics 2009!

OBO Foundry - What


A subset of OBO ontologies whose developers have agreed in advance to accept a common set of principles reec-ng best prac-ce in ontology development designed to ensure
-ght connec-on to the biomedical basic sciences compa-bility interoperability, common rela-ons formal robustness support for logic-based reasoning

Alan Ruttenberg!

Neuroinformatics 2009!

OBO Foundry - How


Collabora-on Peer review Design principles evolve as we learn AHribu-on at all levels Due credit Metrics to measure progress and eec-veness

Alan Ruttenberg!

Neuroinformatics 2009!

Selected OBO Ontologies


Gene Ontology (GO) Rela-on Ontology (RO) Chemical En--es of Biological Interest (CheBI) Phenotype Ontology (PATO) Ontology for Biomedical Inves-ga-ons (OBI) Cell Ontology (CL) Founda-onal Model of Anatomy (FMA) Sequence Ontology (SO) Subcellular Anatomy (SAO)
Neuroinformatics 2009!

Alan Ruttenberg!

Wrap up
Its worth aHemp-ng to construct a seman-c web for neuroscience Seman-c web technologies present opportuni-es for data integra-on at web scale (and corresponding eciencies) There is a growing community of biologists and neuroscien-sts building ontologies, via the OBO Foundry, PONS Please devote some small percentage of your thoughts to the opportuni-es for integra-ng your work into this eort!
Alan Ruttenberg!
Neuroinformatics 2009!