1. The EcoCyc and MetaCyc Pathway/Genome Databases
Peter D. Karp, Ph.D.
Bioinformatics Research Group
SRI International

2. Overview Motivations and terminology Pathway/genome databases
BioCyc collection
EcoCyc, MetaCyc
Pathway Tools software
Bioinformatics Database Warehouse project

3. A E

5. What to do When Theories Become Larger than Minds can Grasp?
Example: E. coli metabolic network
160 pathways involving 744 reactions and 791 substrates
Example: E. coli genetic network
Control by 97 transcription factors of 1174 genes in 630 transcription units
Past solutions:
Partition theories across multiple minds
Encode theories in natural-language text
We cannot compute with theories in those forms
Evaluate theories for consistency with new data: microarrays
Refine theories with respect to new data
Compare theories describing different organisms

6. Solution: Biological Knowledge Bases
Store biological knowledge and theories in computers in a declarative form
Amenable to computational analysis and generative user interfaces
Establish ongoing efforts to curate (maintain, refine, embellish) these knowledge bases
Accepted to store data in computers, but not knowledge
Such knowledge bases are an integral part of the scientific enterprise

7. Pathway Definition A + B C + D A C E
Chemical reactions interconvert chemical compounds
An enzyme is a protein that accelerates chemical reactions
A pathway is a linked set of reactions
Often regulated as a unit
A conceptual unit of cell’s biochemical machine
A + B C + D
A C E

8. Terminology
Model Organism Database (MOD) – DB describing genome and other information about an organism
Pathway/Genome Database (PGDB) – MOD that combines information about
Pathways, reactions, substrates
Enzymes, transporters
Genes, replicons
Transcription factors, promoters, operons, DNA binding sites
BioCyc – Collection of 15 PGDBs at BioCyc.org
EcoCyc, AgroCyc, YeastCyc

9. BioCyc Collection of Pathway/Genome DBs
Computationally Derived Datasets:
Agrobacterium tumefaciens
Caulobacter crescentus
Chlamydia trachomatis
Bacillus subtilis
Helicobacter pylori
Haemophilus influenzae
Mycobacterium tuberculosis RvH37
Mycobacterium tuberculosis CDC1551
Mycoplasma pneumonia
Pseudomonas aeruginosa
Saccharomyces cerevisiae
Treponema pallidum
Vibrio cholerae
Yellow = Open Database
Literature-based Datasets:
MetaCyc
Escherichia coli (EcoCyc)

10. Terminology – Pathway Tools Software
PathoLogic
Prediction of metabolic network from genome
Computational creation of new Pathway/Genome Databases
Pathway/Genome Editors
Distributed curation of PGDBs
Distributed object database system, interactive editing tools
Pathway/Genome Navigator
WWW publishing of PGDBs
Querying, visualization of pathways, chromosomes, operons
Analysis operations
Pathway visualization of gene-expression data
Global comparisons of metabolic networks
Bioinformatics 18:S

11. Pathway Tools Algorithms
Query, visualization and editing tools for these datatypes:
Full Metabolic Map
Paint gene expression data on metabolic network; compare metabolic networks
Pathways
Pathway prediction
Reactions
Balance checker
Compounds
Chemical substructure comparison
Enzymes, Transporters, Transcription Factors
Genes: Blast search
Chromosomes
Operons
Operon prediction

12. Model Organism Databases
DBs that describe the genome and other information about an organism
Every sequenced organism with an active experimental community requires a MOD
Integrate genome data with information about the biochemical and genetic network of the organism
MODs are platforms for global analyses of an organism
Interpret gene expression data in a pathway context
Characterize systems properties of metabolic and genetic networks
Determine consistency of metabolic and transport networks
In silico prediction of essential genes

13. EcoCyc Project – EcoCyc.org
E. coli Encyclopedia
Model-Organism Database for E. coli
Computational symbolic theory of E. coli
Electronic review article for E. coli – over 3500 literature citations
Tracks the evolving annotation of the E. coli genome
Collaborative development via Internet
Karp (SRI) -- Bioinformatics architect
John Ingraham -- Advisor
(SRI) Metabolic pathways
Saier (UCSD) and Paulsen (TIGR)-- Transport
Collado (UNAM)-- Regulation of gene expression
Database content: 18,000 objects

14. EcoCyc = E.coli Dataset + Pathway/Genome Navigator
Pathways: 165
Reactions: 2,760
Compounds: 774
Enzymes: 914
Transporters: 162
Promoters: 812
TransFac Sites: 956
Citations: 3,508
Proteins: 4,273
Transcription
Units:
Factors: 110
Genes: 4,393

15. EcoCyc Procedures All DB updates by 5 staff curators
Information gathered from biomedical literature
Corrections solicited from E. coli researchers
Review-level database
Four releases per year
Available through WWW site, as data files, as downloadable application
Quality assurance of data and software:
Evaluate database consistency constraints
Perform element balancing of reactions
Run other checking programs
Display every DB object

16. MetaCyc: Metabolic Encyclopedia
Nonredundant metabolic pathway database
Describe a representative sample of every experimentally determined metabolic pathway
Literature-based DB with extensive references and commentary
Pathways, reactions, enzymes, substrates
460 pathways, 1267 enzymes, 4294 reactions
172 E. coli pathways, citations
Nucleic Acids Research 30:
Jointly developed by SRI and Carnegie Institution
New focus on plant pathways

17. Family of Pathway/Genome Databases
MetaCyc

18. Pathway Tools Implementation Details
Allegro Common Lisp
Sun and PC platforms
Ocelot object database
250,000 lines of code
Lisp-based WWW server at BioCyc.org
Manages 15 PGDBs

19. Pathway Tools Architecture
Genome
Navigator
WWW
Server
X-Windows
Graphics
Object Editor
Pathway Editor
Reaction Editor
GFP API
Oracle
Object DBMS

20. Ocelot Knowledge Server Architecture
Frame data model
Classes, instances, inheritance
Frames have slots that define their properties, attributes, relationships
A slot has one or more values
Each value can be any Lisp datatype
Slotunits define metadata about slots:
Domain, range, inverse
Collection type, number of values, value constraints
Transaction logging facility
Schema evolution

21. Ocelot Storage System Architecture
Persistent storage via disk files, Oracle DBMS
Concurrent development: Oracle
Single-user development: disk files
Read-only delivery: bundle data into binary program
Oracle storage
DBMS is submerged within Ocelot, invisible to users
Relational schema is domain independent, supports multiple KBs simultaneously
Frames transferred from DBMS to Ocelot
On demand
By background prefetcher
Memory cache
Persistent disk cache to speed performance via Internet

22. The Common Lisp Programming Environment
Gatt studied Lisp and Java implementation of 16 programs by 14 programmers (Intelligence 11: )

23. EcoCyc WWW Server

24. Pathway/Genome DBs Created by External Users
Plasmodium falciparum, Stanford University
plasmocyc.stanford.edu
Mycobacterium tuberculosis, Stanford University
BioCyc.org
Arabidopsis thaliana and Synechosistis, Carnegie Institution of Washington
Arabidopsis.org:1555
Methanococcus janaschii, EBI
Maine.ebi.ac.uk:1555
Other PGDBs in progress by 24 other users
Software freely available
Each PGDB owned by its creator

25. Global Consistency Checking of Biochemical Network
Given:
A PGDB for an organism
A set of initial metabolites
Infer:
What set of products can be synthesized by the small-molecule metabolism of the organism
Can known growth medium yield known essential compounds?
Pacific Symposium on Biocomputing p

26. Algorithm: Forward Propagation
Nutrient
set
Products
Transport
PGDB
reaction
pool
Metabolite
set
“Fire”
reactions
Reactants

27. Results Phase I: Forward propagation
21 initial compounds yielded only half of 38 essential compounds for E. coli
Phase II: Manually identify
Bugs in EcoCyc (e.g., two objects for tryptophan)
Missing initial protein substrates (e.g., ACP)
Missing pathways in EcoCyc
Phase III: Forward propagation with 11 more initial metabolites
Yielded all 38 essential compounds

28. Nutrient-Related Analysis: Validation of the EcoCyc Database
Results on EcoCyc:
Phase I:
Essential compounds
produced 19
not produced 19
Total compounds
produced: (28%)
Reactions
Fired (31%)

29. Missing Essential Compounds Due To
Bugs in EcoCyc
Narrow conceptualization of the problem
Protein substrates
Incomplete biochemical knowledge

30. Nutrient-Related Analysis: Validation of the EcoCyc Database
Results on EcoCyc:
Phase II (After adding 11 extra metabolites):
Essential compounds
produced 38
not produced 0
Total compounds
produced: (49%)
not produced: (51%)
Reactions
Fired (58%)
Not fired (42%)

31. Pathway Tools Misconceptions
PathoLogic
Does not re-annotate genomes
Pathway Tools does not handle quantitative information
Pathway/Genome Editors do not work through the web

32. HumanCyc: Human Metabolic Pathway Database Consortium
Construct DB of human metabolic pathways using PathoLogic
Link to human genome web sites
Hire one curator to refine and curate with respect to literature over a 2 year period
Remove false-positive predictions
Insert known pathways missed by PathoLogic
Add comments and citations from pathways and enzymes to the literature
Add enzyme activators, inhibitors, cofactors, tissue information
Available as flatfiles and with Pathway/Genome Navigator
New versions to be released every 6 months

33. Summary Pathway/Genome Databases
MetaCyc non-redundant DB of literature-derived pathways
14 organism-specific PGDBs available through SRI at BioCyc.org
Computational theories of biochemical machinery
Pathway Tools software
Extract pathways from genomes
Morph annotated genome into structured ontology
Distributed curation tools for MODs
Query, visualization, WWW publishing

34. BioCyc and Pathway Tools Availability
WWW BioCyc freely available to all
BioCyc.org
Six BioCyc DBs openly available to all
BioCyc DBs freely available to non-profits
Flatfiles downloadable from BioCyc.org
Binary executable:
Sun UltraSparc-170 w/ 64MB memory
PC, 400MHz CPU, 64MB memory, Windows-98 or newer
PerlCyc API
Pathway Tools freely available to non-profits

35. BioCyc.org Acknowledgements Funding sources:
SRI
Suzanne Paley, Pedro Romero, John Pick, Cindy Krieger, Martha Arnaud
EcoCyc Project
Julio Collado-Vides, Ian Paulsen, Monica Riley, Milton Saier
MetaCyc Project
Sue Rhee, Lukas Mueller, Peifen Zhang, Chris Somerville
Stanford
Gary Schoolnik, Harley McAdams, Lucy Shapiro, Russ Altman, Iwei Yeh
Funding sources:
NIH National Center for Research Resources
NIH National Institute of General Medical Sciences
NIH National Human Genome Research Institute
Department of Energy Microbial Cell Project
DARPA BioSpice, UPC
BioCyc.org