Presentation is loading. Please wait.

Presentation is loading. Please wait.

Pathway Knowledge Base: A Public Repository for Searching Biological Pathways Nikesh Kotecha 1, Kyle Bruck 1, William Lu 1 and Nigam Shah 1 1 Department.

Similar presentations


Presentation on theme: "Pathway Knowledge Base: A Public Repository for Searching Biological Pathways Nikesh Kotecha 1, Kyle Bruck 1, William Lu 1 and Nigam Shah 1 1 Department."— Presentation transcript:

1 Pathway Knowledge Base: A Public Repository for Searching Biological Pathways Nikesh Kotecha 1, Kyle Bruck 1, William Lu 1 and Nigam Shah 1 1 Department of Biomedical Informatics, Stanford University AB C

2 PKB C BA Outline The problem with pathways…The problem with pathways… BioPAX and data exchangeBioPAX and data exchange Pathway Knowledge BasePathway Knowledge Base –Querying over multiple data sources –Data import and query process EvaluationEvaluation Conclusion & Next StepsConclusion & Next Steps

3 PKB C BA Pathways convey biological phenomena Sundaraaj et. al, 2005

4 PKB C BA Users interpret non-standard pathway diagrams Insulin signaling pathways – KEGG & BioCarta

5 PKB C BA The problem with pathways… Too many sources to investigateToo many sources to investigate –>200 databases in the Pathway Resource List –http://www.pathguide.org Pathway diagrams are the primary way for consuming informationPathway diagrams are the primary way for consuming information –Non-standard –Incorporates a handful of genes and proteins –Difficult to compute over “Pathway” representation is not easy to define“Pathway” representation is not easy to define –Appropriate levels of abstraction –Standard naming systems for molecules –Appropriate numbers of components

6 PKB C BA BioPAX is an emerging standard for pathway representation

7 PKB C BA BioPAX allows for creation of a central resource BioPAXBioPAX –Standardized representation of Pathway data in OWL Provide method for exchange that promotes interoperability between known data sources.Provide method for exchange that promotes interoperability between known data sources. We use BioPAX level 2 to integrate Kegg, BioCyc and Reactome

8 PKB C BA Pathway Knowledge Base (PKB) Infrastructure to integrate pathway information from multiple sources into a central resourceInfrastructure to integrate pathway information from multiple sources into a central resource –BioPAX and Oracle RDF model Methods to query pathways in these databasesMethods to query pathways in these databases –SQL/SPARQL A web interface that allows users and other programs to query and access these pathwaysA web interface that allows users and other programs to query and access these pathways –JSP/AJAX

9 PKB C BA Search pathways across data sources and species

10 PKB C BA Visualize and navigate pathway results

11 PKB C BA Custom queries enabled via advanced search

12 PKB C BA Under the Hood…

13 PKB C BA Architecture Pathway Viewer tools Patika, Cytoscape etc. Web Client Interface RDF ObjectsPKB C BA BioPAX files Reactome BioCyc BioPAX Import Object Web Server Search Pathway BioPAX output SQL HTTP JSP JENA

14 PKB C BA Sample Query: List all reactions in PKB select distinct t,x,y FROM ( select 'HUMAN' as t,x,y from TABLE(SDO_RDF_MATCH('(?y rdf:type :biochemicalReaction) (?y :NAME ?x)', SDO_RDF_Models('human'), null,SDO_RDF_Aliases(SDO_RDF_Alias('','http://www.biopax.org/rele ase/biopax-level2.owl#')),null)) UNION ALL select 'YEAST' as t,x,y from TABLE(SDO_RDF_MATCH('(?y rdf:type :biochemicalReaction) (?y :NAME ?x)', SDO_RDF_Models('yeast'), null,SDO_RDF_Aliases(SDO_RDF_Alias('','http://www.biopax.org/rele ase/biopax-level2.owl#')),null)) UNION ALL select 'ECOLI' as t,x,y from TABLE(SDO_RDF_MATCH('(?y rdf:type :biochemicalReaction) (?y :NAME ?x)', SDO_RDF_Models('ecoli'), null,SDO_RDF_Aliases(SDO_RDF_Alias('','http://www.biopax.org/rele ase/biopax-level2.owl#')),null)) )ORDER BY x Each species has its own RDF model Query across species via UNION ALL Determine data source via URI: 1HUMANRecruitment of elongation factors to form HIV-1 elongation complex _factors_to_form_HIV_1_elongation_complex

15 PKB C BA Adding new species and data sources Add new speciesAdd new species –Create TABLE for storing rdf data –Create an RDF_MODEL for that species –Insert data as N-Triples Add new data sourceAdd new data source –Obtain BioPAX pathway data –Convert BioPAX OWL files into the triples format using JENA –Update all unique identifiers to have a custom URI* (http://pkb.stanford.edu/biopax#datasourcename) –Upgrade biopax identifiers from level 1 to level 2* –Insert data into table

16 PKB C BA Evaluation

17 PKB C BA PKB Query Performance HumanE. coliYeastTotal # of pathways # of reactions # of triples ,131,479 Summary of data available in PKB Example query times Queries run on a Dell PowerEdge 2800 w/ 4 GB RAM and Oracle QueryTime (s) Search for pathway titles containing phospholipase1.3 Find all proteins with bcl1 in their name or synonym16.1 Find all pathways with protein bcl135.3 List all entities to the left or right of hydroxyanthranilate39.7

18 PKB C BA Certain queries are easier with relational structures Pathway APathway B Reaction D Protein E Reaction C PathwayReactionProtein ACE BDE Query: Given protein E, find all pathways Relational Structure Directed Graph

19 PKB C BA Pathway Knowledge Base BenefitsBenefits –Centralized resource for querying pathway information across data sources and species –Demonstrates use of BioPAX and RDF for exchange and integration of pathway data LimitationsLimitations –Storing data only as RDF graph structures limits query performance –Import requires manipulation of namespaces –Querying options and methods are based on SPARQL but do not currently support all the features specified in SPARQL

20 PKB C BA Next Steps in PKB? Pathway differences between data sourcesPathway differences between data sources –How does the galactose metabolism differ b/w ecocyc and reactome Pathway mergingPathway merging –Merge information from the same pathways from different data sources Pathway consistency checkingPathway consistency checking –BioPAX rules for testing pathway models

21 PKB C BA Acknowledgements PKB TeamPKB Team –Kyle Bruck –William Lu –Nigam Shah FeedbackFeedback –Daniel Rubin –Russ Altman HostingHosting –National Center for Biomedical Ontologies Funding SourcesFunding Sources –Stanford Medical Informatics –National Library of Medicine –National Institute of Health

22 PKB C BA Thanks. Any Questions?


Download ppt "Pathway Knowledge Base: A Public Repository for Searching Biological Pathways Nikesh Kotecha 1, Kyle Bruck 1, William Lu 1 and Nigam Shah 1 1 Department."

Similar presentations


Ads by Google