1. Representing Biological Processes: The Reactome Database Gopal Gopinathrao 1 & Peter D’Eustachio 1,2 1 Cold Spring Harbor Laboratory 2 NYU School of Medicine gopinath@cshl.edu deustp01@med.nyu.edu

2. Reactome is - reductionist. All of biology can be represented as events that convert input physical entities into output physical entities. - a generic parts list. Tissue and state specificity of events are not captured. - qualitative. Kinetic parameters and data are not captured. - human-centric. Experiments can use reagents from diverse sources, but most biological processes take place in single species, and our focus is on human biological processes. - manually curated. Events are annotated by expert curators, and linked to published data. - open source. All data and software are freely downloadable and reusable.

3. Data model in a nutshell Pathway Reaction CatalystActivity Output 1 Reaction Input 1 Input 2 Output 2 Regulation

4. Annotating more details - post-translational modifications of proteins - exact locations of entities and events Annotating more ambiguities - sets of entities - defined, open, and candidate - incompletely specified entities - “black box” reactions

5. A geometrical compartment set for locating molecules in human cells

6. Hemo- stasis Apop- tosis Insulin signal- ing Notch signal- ing Glucagon signaling Cell cycle & DNA replication DNA repair Transcription Translation Posttransla- tional modifi- cations TCA cycle Lipid metabolism Amino acid metabolism Nucleotide metabolism Xenobiotic metabolism Carbohydrate metabolism The starry sky view of all of Reactome HIV & Influenza life cycles Sterol metab- olism

7. Reactome Home Page http://brie8.cshl.edu/cgi-bin/frontpage?DB=gk_central

8. Reactome Event Page http://brie8.cshl.edu/cgi-bin/eventbrowser?DB=gk_central&ID=163767&

9. Export Formats BioPAX pathway converted from "DNA Replication" in the Reactome database. DNA Replication Studies in the past decade have suggested that the basic mechanism of DNA replication initiation is conserved in all kingdoms of life. Initiation in unicellular eukaryotes, in particular Saccharomyces cerevisiae (budding yeast), is well

10. Bioinformatics Access BioMart API MySQL/Perl API MySQL/Java API SOAP/WSDL Interface (multiple languages) Flat files Database dumps Local site install (instructions going into CPBI)

11. Inference Statistics direct curation underway

12. Validation of inference Comparison of manually curated yeast reactions from YBP with inferred reactions from human Reactome Sensitivity: 72% Specificity: 78%

13. Inferring chicken reactions from curated human ones

15. Gaps in Reactome Gopal Gopinathrao, PhD Reactome, CSHL

16. 1) Gaps in Reactome annotation 2) Gaps in annotate-able information 3) What a network / pathway ontology can do to fill this gap?

17. Information Cellular PathogensPathogens

18. PathogensPathogens Information Metabolism

19. Signaling Information

20. Protozoan/Host interactions Developmental pathways Transcriptional regulation Feedback loops Neuroscience topics Degenerative diseases Synaptic processes Cancer processes OMIM-functional (biochemical) Complex diseases Cellular differentiation, Regulation Domains of Biology waiting to be Reactomized

21. Metabolism Cellular housekeeping Information Signaling Pathogens/ Host interactions 476 755 376 414 600 Unique human proteins used in pathways (in March 2008) 2500 Swissprot section of UniProt ~16,000

22. 0 1000 2000 3000 4000 5000 6000 10203040 release total proteins unique proteins unique + isoforms Gaps in Reactome annotation

24. Some pathway/Int dbs are more equal?

25. Are all Swissprot proteins annotatable for pathways/interactions? Can all interactions can be placed in a biologically relevant ‘pathway’ or even sub-graphs of a network? If yes, who is going to validate and how, the biological ‘truth’ of any subgraphs derived from a network? [Terms of biological truth - tissue, regulation, developmental stage, expression …] Mind what gets filled in…

26. Watching the gap… Adding in pathway data decomposed to interactions … Adding PPI data to the above …

27. NBC Predictions in Reactome

28. How a network / pathway ontology may help to fill the gap in pathway annotations..

29. A B C B A D C D C A A<-----| B Known Novel New regulatory event A+B+C+D Feedback loop ABCD complex

30. 1. A+B+C+D 2.Interaction of C and D may regulate ABCD complex formation Updated model for curation would be: C D A B Novel Feedback loop 3. Post-translational inhibition of B by A may result in down regulation of A, there by affecting the stability of complex ABCD Evidence from a network ontology A<-----| B New regulatory event Evidence from a network ontology in a model organism

31. The Team CSHL –Lincoln Stein (PI) –Gopal Gopinathrao (managing editor) –Marc Gillespie, Lisa Matthews, Bruce May, Mike Caudy (curators) –Guanming Wu, Alex Kanapin (developer) EBI –Ewan Birney (coPI) –Esther Schmidt, Imre Vastrik, David Croft (developers) –Bernard de Bono, Bijay Jassal, Phani Garapati (curators) NYU –Peter D’Eustachio (co-PI; editor-in-chief) –Shahana Mahajan (curator) P41 HG003751