(last update 09/07/2007)
To verify that we obtained a more detailed and systematic pathway representation with respect to available information than in existing pathway databases, we chose to compare it to Reactome database. Reactome database was chosen following these three criteria: it is a publicly available pathway database containing rich and regularly updated information about signalling cascades and cell cycle in human cells; it represents comparable level of interaction details as our pathway representation; and it provides its up-to-date content in the standard BioPAX format.
Our RB pathway in the BioPAX format was used as a query to the BioPAX dump (dated on 15 March 2007) of Reactome database (see Methods). As a result, we extracted a sub-network of the whole Reactome network involving proteins that are common with our pathway. The obtained network is therefore the intersection between proteins and reactions found in our pathway and the same set of proteins searched in Reactome and that are involved in all possible reactions.
At the level of physical entities, the query engine managed to identify 49 proteins and 40 protein complexes against 78 and 94, respectively, contained in our RB map (the identification of entities was made based on both protein names and Uniprot identifiers). Some difficulties were found due to the fact that an entity in Reactome can be a group of proteins such as “CyclinA/Cdk1 substrate proteins”. Counting them, one finds 77 related entities in Reactome. Information about proteins such as E2F6, E2F8, EPC1, or DP2 and many others was not found in Reactome database.
At the level of distinct species, one can identify 105 species corresponding to 77 entities found in Reactome against 140 species (for the same set of proteins) in our RB map. Note that in Reactome the same species might appear under different names (corresponding to different “generic” entities).
At the level of reactions, one can identify in Reactome 64 reactions connecting the identified species against 135 reactions (including reactions of degradation) in RB map, corresponding to the same proteins found in Reactome.
All this ensured us that the representation of RB network which we provide in this paper is more systematic and focused than that contained in the general purpose Reactome pathway database. However, there are few details contained in Reactome that we decided not to put in our description of RB pathway such as the mechanisms and proteins regulating DNA repair.
Reaction density = number of reactions / number of proteins = 64/49 = 1.31 (in our diagram 2.5)
- Comparison with Transpath
The method for comparing our pathway to Transpath was similar to that of Reactome. The results are the following:
- Entities (proteins or protein groups):
53 proteins in TRANSPATH, 78 in our diagram
28 complexes in Transpath, 68 in our diagram (94 total)
- Species:
145 in TRANSPATH, 140 in our diagram
- Reactions:
106 (incl. 35 with p53) in TRANSPATH, 135 in our diagram
Reaction density = 106/53 = 2.0 (in our diagram: 1.73)
