Biology Atlas

Map / Concepts

concept · intermediate · partly written

How a pathway is encoded: BioPAX, SBML, GO-CAM and Reactome's own model

A pathway is not one thing stored two ways: BioPAX keeps entities, hierarchy and order, SBML a runnable skeleton, GO-CAM causal activities, and each conversion drops something.

Summary

On one Reactome pathway of 16 reactions, the SBML export kept all 16 reactions but flattened nested complexes, dropped the pathway hierarchy and step order, and carried no rate laws, while the BioPAX export kept them. The one measured Reactome-to-GO-CAM conversion gave 3,593 complete activity units from 11,570 reactions, by the converter's own authors. 12

Reported, not yet replicated

What each format is more

BioPAX is a community OWL vocabulary for exchanging qualitative pathway knowledge, Level 3 released July 2010, a defined standard. SBML is an XML format for computational models that a simulator can run; the specification read is Level 3 Version 2 Core Release 2, and Reactome's export is Level 3 Version 1. GO-CAM links several Gene Ontology annotations into one causal model and is one consortium's schema. Reactome's data model is a database schema from which the BioPAX and SBML files are exported; one group's choice. 345

Reported, not yet replicated

One pathway opened in both exports more

Lysine catabolism (Reactome R-HSA-71064) lists 14 direct events plus a sub-pathway of 3 reactions. Its BioPAX level 3 file (Reactome release 97) has 16 reactions, 2 Pathway objects with 17 pathway steps, 10 nested complexes, 14 catalysis objects and 18 stoichiometry objects. Its SBML level 3 version 1 file (release 98) has 16 reactions, all irreversible, in one flat list with no groups, 6 flat complexes carrying a note that Reactome's nested complexes cannot be fully represented in that level, 14 modifier references, and no kinetic laws. The files are from two releases, so small differences are unresolved. In a single-reaction case, CRYM (crystallin mu, HGNC:2418) reducing P2C to PPCA in Reactome's abbreviations (reaction R-HSA-5693347), the two files agree on chemistry and differ in typing: an explicit catalysis object in BioPAX against an untyped modifier in SBML. 16

Reported, not yet replicated

Reactome to GO-CAM, as measured by the converter's authors more

From Reactome v73, 11,570 human reactions were converted after disease and drug reactions were removed; 3,593 came out as complete activity units and 7,986 as incomplete. Reasons printed: Reactome treats the participants of binding events as equals while a GO-CAM activity needs one enabling gene product (4,158 binding and 416 dissociation events); 3,529 incomplete conversions got no location because a reaction spans several compartments and a GO-CAM activity may have only one; 5,572 reactions were linked only to the root biological-process term because their Reactome parent pathway has no specific GO term. Of 11,753 reaction-to-reaction links, 5,435 became causally upstream of, 5,273 directly provides input for, 878 directly positively regulates and 140 directly negatively regulates, so most links carry order and no sign. Authors include Reactome and Gene Ontology people, so this is not an outside audit. Example genes in the paper: KRAS (KRAS proto-oncogene, GTPase, HGNC:6407) and NOTCH3 (notch receptor 3, HGNC:7883). 26

Reported, not yet replicated

The same format used differently more

Beust and colleagues, independent of Reactome and the Gene Ontology Consortium, compared the BioPAX exports of five databases and found high heterogeneity: only Reactome and PathBank use the pathway step class, protein mapping to UniProtKB exceeded 85% for two Reactome versions, HumanCyc and PathBank, and small-molecule mapping to ChEBI was above 75% for three of the databases and about 70% for the rest. Pathway Commons, which merges 22 databases as BioPAX, says much effort was needed to tune providers' output. 78

Reported, not yet replicated

BioPAX to SBML elsewhere more

A relation such as A modulates B is not a reaction and SBML core has only reactions; older converters either converted such relations to reactions or dropped them, and Büchel and colleagues used the SBML qual package for the NCI Pathway Interaction Database without reporting counts. Of over 140,000 Path2Models models built from pathway maps, 6,204 reactions (0.22%) got a rate law from SABIO-RK and the rest got generated laws with new parameters set to 1.0. No round-trip test on a whole database was found. 910

Reported, not yet replicated

How far GO-CAM has reached more

The Gene Ontology Consortium reports 1,571 expert-curated GO-CAM pathways as of July 2025, and says the Reactome-derived GO-CAMs are not yet public. 11

Reported, not yet replicated

Orthology as a route in more

Human Reactome-derived GO-CAMs were converted to mouse ones by orthology: for glycolysis, gluconeogenesis and pyruvate metabolism, 44 of 60 human gene products had high-confidence mouse orthologs, and of the mouse assertions 43 had direct mouse experimental evidence and 16 rested on sequence orthology to human or rat. Gene products and assertions are different units. 12

Reported, not yet replicated

Plain answer

A pathway format fixes what counts as one entity and one step: BioPAX keeps physical entities, hierarchy and order; SBML keeps a quantity model; GO-CAM keeps one gene product doing one function in one place. Moving between them drops structure. 342

Reported, not yet replicated

Why it matters

Amit asked how protein pathways and chemical interactions are represented; the format decides what a mechanism can say about direction, stoichiometry and location. 2

Our inference

Real numbers

Reactome to GO-CAM conversion 2026-11-18

11,570 reactions converted; 3,593 complete, 7,986 incomplete; 3,529 incomplete ones lacked a location and 5,572 reactions got only the root process term. 2

Reported, not yet replicated

GO-CAM pathways curated 2026-11-18

1,571 expert-curated GO-CAM pathways as of July 2025. 11

Reported, not yet replicated

Still open here

Not yet written: Equation.

Linked from

Sources

  1. Reactome exports of R-HSA-71064 (lysine catabolism) in BioPAX level 3 and SBML level 3 version 1, and of R-HSA-5693347.. database, not peer reviewed; opened 2026-11-18 all records citing this
  2. Good BM, Van Auken K, Hill DP, Mi H, et al.. Reactome and the Gene Ontology: digital convergence of data resources. Bioinformatics (2021). paper; opened 2026-11-18 all records citing this
  3. Demir E, Cary MP, Paley S, Fukuda K, et al.. The BioPAX community standard for pathway data sharing. Nat Biotechnol (2010). paper; opened 2026-11-18 all records citing this
  4. Hucka M, Bergmann FT, Chaouiya C, Dräger A, et al.. The Systems Biology Markup Language (SBML): Language Specification for Level 3 Version 2 Core Release 2. J Integr Bioinform (2019). paper; opened 2026-11-18 all records citing this
  5. Data Model - Reactome Pathway Database.. organisation-page, not peer reviewed; opened 2026-11-18 all records citing this
  6. HGNC REST records for IL2RA, CCR7, MS4A1, CD28, CRYM, KRAS and NOTCH3. HGNC. database, not peer reviewed; opened 2026-11-18 all records citing this
  7. Beust C, Becker E, Théret N, Dameron O. BioPAX in 2024: Where we are and where we are heading. Comput Struct Biotechnol J (2024). paper; opened 2026-11-18 all records citing this
  8. Rodchenkov I, Babur O, Luna A, Aksoy BA, et al.. Pathway Commons 2019 Update: integration, analysis and exploration of pathway data. Nucleic Acids Res (2020). paper; opened 2026-11-18 all records citing this
  9. Büchel F, Wrzodek C, Wrzodek C, Mittag F, et al.. Qualitative translation of relations from BioPAX to SBML qual. Bioinformatics (2012). paper; opened 2026-11-18 all records citing this
  10. Büchel F, Rodriguez N, Swainston N, Wrzodek C, et al.. Path2Models: large-scale generation of computational models from biochemical pathway maps. BMC Syst Biol (2013). paper; opened 2026-11-18 all records citing this
  11. Gene Ontology Consortium. The Gene Ontology knowledgebase in 2026. Nucleic acids research (2026). paper; opened 2026-11-18 all records citing this
  12. Hill DP, Drabkin HJ, Smith CL, Van Auken KM, et al.. Biochemical pathways represented by Gene Ontology-Causal Activity Models identify distinct phenotypes resulting from mutations in pathways. Genetics (2023). paper; opened 2026-11-18 all records citing this
Where it sits on the map
History of this record (1)

This record has been changed 1 time, on the dates below. The note kept against each change is written for the people who maintain the atlas, in their own shorthand, so it is not shown here; it is in the record itself.

Status
partly written
Last updated
2026-11-18
Record
content/concepts/pathway-exchange-formats.json