Integrate SABIO-RK (human) data

[IgorRodchenkov]

Try SABIO-RK (human) data (BioPAX)

[IgorRodchenkov]

Home page: http://sabio.h-its.org Download data (trying): http://sabiork.h-its.org/sabioRestWebServices/searchKineticLaws/biopax?q=Organism:"Homo%20sapiens" (awaiting response... never ends)

Cannot submit a question using the feedback/contact form at: http://sabio.h-its.org/contactFormSabio

[IgorRodchenkov]

Dear Igor,

thank you very much for your interest in SabioRK and for proposing to integrate SabioRK into Pathway Commons.

By perfmoring the corresponding query "http://sabiork.h-its.org/sabioRestWebServices/searchKineticLaws/biopax?q=Organism:%22Homo%20sapiens%22" we noticed that there was an error message thrown during the export process due to some inconsistencies in our database which were fixed in the meantime. However, by using your group's online BioPAX evaluation toolt, it seems that there are errors (eg on cardinality) in the current output format as you may also have been noticed.

I attached a file with the exported entities for Organism="Homo sapiens" in SBML format (level 2, version 4). Using this file, you should be able to generate the corresponding .owl file in BioPAX format by using your SBML2BioPAX tool.

Thank you also for offering your help in fixing issues in regard to BioPAX export. However, at the present the datasets stored in our database are not freely available and therefore it makes not much sense to release the software used for querying without the underlying datastructure (and data).

Actually the export is performed by translating / mapping SabioRK objects directly to the corresponding BioPax graph; there is no intermediate SBML export step. Thus we (actually I) have to implement any changes, bug fixes and improvements.

Best regards, Andreas Weidemann

[IgorRodchenkov]

@emekdemir @jvwong @cannin @ozgunbabur

A.f.a.i.k., we have no quality "SBML2BioPAX tool", do we, anyone?

There are

• https://www.ebi.ac.uk/biomodels/tools/converters/ (online service - does not work for these data - returns a very large mixed/messed XML/HTML result file with extension ".input"...)
• http://sbfc.sourceforge.net/mediawiki/index.php/Main_Page (the converter to L3 does not look good enough: it does not handle RDF in SBML annotations properly; does not produce any EntityReferences, BioSource; generates dangling empty entities, e.g., interactions; adds e.g., Uniprot unification xrefs to Proteins directly; uses 'memberPhysicalEntity' property, which should be better avoided; etc... - that's what I saw at a glance).

Mapping SBML to BioPAX Level2 there makes more sense to me compared with mapping to BioPAX Level3 (where EntityReferences should be used, etc.). So, I will try the following two-step conversion:

1. mapping the origibal SBML data to BioPAX L2 with the sbfc tool;
2. converting the above result to BioPAX L3 with Paxtools console 'toLevel3' command (- official BioPAX L2 to L3 model converter).

...

[IgorRodchenkov]

@gbader @ozgunbabur

I converted the data to BioPAX L2 with the sbfc tool (@armish seems to develop it in 2015), and then - to L3 with Paxtools (which I am sure would normally produce good model if input L2 model were good; though it does in fact mess-up original URIs due to re-using some entities, and it e.g. removes Stoichiometry objects having coeff<2.0).

Looks, there are many semantic problems, besides empty interactions, such as - e.g., the following reaction

<bp:BiochemicalReaction rdf:about="conversion_REAC_8179">
 <bp:xref rdf:resource="http://identifiers.org/sabiork.reaction/13245" />
 <bp:xref rdf:resource="http://identifiers.org/ec-code/3.2.2.27" />
 <bp:xref rdf:resource="http://identifiers.org/taxonomy/9606" />
 <bp:right rdf:resource="RIGHT_1_conversion_REAC_8733_SPC_61_Cell" />
 <bp:right rdf:resource="RIGHT_1_conversion_REAC_8179_SPC_6655_Cell" />
 <bp:left rdf:resource="RIGHT_1_conversion_REAC_2337_SPC_1770_Cell" />
 <bp:left rdf:resource="LEFT_0_conversion_REAC_1960_SPC_40_Cell" />
</bp:BiochemicalReaction>

literally means: someDnaRegion + H2O => anotherDnaRegion + Uracil , where someDnaRegion is KEGG C00039; anotherDnaRegion is C03484 (not found in KEGG db; but in the sbfc it was probably meant to mean "Apyrimidinic site in DNA"). There is no Catalysis or Control interaction, controlling this reaction, where EC:3.2.2.27 (uracil-DNA glycosylase) would probably play controller role (instead of just using Xref to EC:3.2.2.27).