The knowledge-driven exploration of integrated

biomedical knowledge sources facilitates the

generation of new hypotheses

Vinh Nguyen, Olivier Bodenreider, Todd Mining, AmitSheth

Context

• Chagas disease

– Infectious disease caused by Trypanosomacruzi(T. cruzi) parasite

– Affects 8-10 million people in South America, about 20,000 deaths per year

– Finding drug targets and vaccine

210/21/2011

Context

310/21/2011

Why integrate two?

• Synergistically exploit the two sources

– If literature indicates certain finding (for the same or other organism [mice, rat], for same or different process [e.g. for gene expression], does my data support the same finding?

– I find a pattern in my data, or a pathway component – is this reported in literature by someone in the same or similar context?

10/21/2011 4

10/21/2011 5

The knowledge-driven integrated exploitation of complementary sources, which can facilitate generation of new hypotheses for additional investigations and insights.

Use case

• Data sources

– Text data from PubMed

– Experimental data about Chagas disease

• Objectives/capabilities

– Query experimental and literature (text) together

– Find new insights into Chagas disease

610/21/2011

Background Knowledge

• Two knowledge bases were linked/partly integrated– The Biomedical Knowledge Repository (BKR) [NLM]

– The Parasite Knowledge Repository for Chagas disease (PKR)

710/21/2011

Knowledge source - BKR

• Created at NLM by Dr. Olivier Bodenreider and Dr. Thomas Rindflesch

• ~ 20 million predications from 6 million PubMed articles by SemRep

• Example

810/21/2011

Knowledge source - PKR

Created by T.cruziproject at Kno.e.sis, WSU with CTEGD, UGA primarily for semantic analysis of parasite experimental data from/including:

• The expression level of T. cruzi genes (from Microarray analysis data)

• The presence of proteins encoded by T. cruzi genes (using Proteome analysis)

• Ontologies for annotation– Parasite Experiment Ontology (PEO)– Parasite Lifecycle Ontology (PLO)

910/21/2011

Linking the two knowledge bases

• Orthology relation

– Gene function is usually conserved across species

1010/21/2011

How to explore the integrated knowledge bases?

• Knowledge-driven exploration process

• Two steps of the process– Navigate the chain of named relationships

• Graph expansion• Graph restriction

– Interpret the chain to infer new hypothesis

1110/21/2011

Example of Exploration process• Sample chain

1210/21/2011

Exploration process – Demo

iExplorevideo demo is available at

http://www.youtube.com/watch?v=-7-BgBDTSjc

1310/21/2011

Exploration process -Interpretation

• Sample chain

1410/21/2011

The T. cruzi orthologs of human genes inhibited by itraconazole may also be inhibited by itraconazole and thus would be candidates for further studies into the mechanism(s) of action of the drug itraconazole on T. cruzi

-> Hypothesis

Significance

• Biologically

– Biologists follow the exploration process to generate novel hypotheses

– Can easily be generalized to other diseases

• Technically

– Technical details (e.g., SPARQL queries) hidden from the user

– Scalable over millions of predications

10/21/2011 15

Technologies

• Semantic Web

– RDF, SPARQL standards

– Virtuoso triple store

• iExplore development

– Client side: Flex application

– Server side: Java

10/21/2011 16

Conclusion• Integrate knowledge bases

– BKR

– PKR

• Knowledge-driven exploration

– Two-step process: navigation and interpretation

– Assist biologists to explore the integrated knowledge bases

1710/21/2011

Future work• Validation of novel hypotheses

– Collaborating with UGA researchers in the biomedical aspect

– Incorporate the validation into iExplore

• Automate the interpretation step

– Learning the way biologists use their background knowledge to interpret the chain of named relationship

1810/21/2011

Acknowledgements

Dr. Thomas Rindflesch

May Cheh

Dr. Clement McDonald

Dr. BastienRance

Jonathan Mortensen

John Nguyen

Thang Nguyen

Lister Hill Center

1910/21/2011

Dr. CarticRamakrishnan

Dr. Priti Parikh

Joshua Dotson

SarasiLalithsena

10/21/2011 20

10/21/2011 21

Further information, please visit http://knoesis.wright.edu/iExplore