Version 16 (modified by ogishima, 14 years ago)

--

SystemsBiologySBML

In systems biology field, most curated models are developed by for example  CellDesigner, and are described in  SBML format. Each model is publised in each paper, and in each SBML file. To integrate published models, semantic web approach is very useful. BioPAX provides a common exchange format for biological pathways and models in OWL.  http://www.biopax.org/

Goal

Goal is first of all an integration of curated models, in other words, pathways. As described above, each pathway is publised in each paper, and in each SBML file. However, as you know, pathways are closely related each other. If we can integrate these curated pathways, we can overview pathways and search pathways to find novel hypothesis. It will be very helpful for us to obtain novel insight and hypothesis.

Usecase - Alzheimer disease pathway

One usecase is Alzheimer disease study. We has developed a curated pathway database for Alzheimer disease named AlzPathway?. AlzPathway? collects signaling and metabolic pathways relevant to Alzheimer disease. By only AlzPathway?, we already discovered novel hypothesis of relationships between amyloid beta and tau phosphorylation pathways which has been a missing link before.

Importantly, pathways relevant to Alzheimer disease are not exclusively relevant to Alzheimer disease. For example, Alzheimer disease pathway is closely related to apoptosis pathway, and if someone published a curated model for detailed apoptosis, an integration of AlzPathway? and apoptosis pathway will be very helpful for researchers in Alzheimer disease to obtain novel insight and hypothesis about relationships between Alzheimer disease and apoptosis.

Datasets

Convertor

BioPAX Level 3

No convertor now.

BioPAX Level 2

The convertor sbml2biopax from SBML to BioPAX OWL is provided by BioModels? Compneur Group.

The sbml2biopax requires  libSBML, and it works with the latest libSBML (libsbml.4.0.1).

$ wget http://www.ebi.ac.uk/compneur-srv/sbml/convertors/sbml2biopax-1.3.tar.gz
$ tar xvfz sbml2biopax-1.3.tar.gz; cd sbml2biopax-1.3; cp ~/alzpathway.xml .
$ sh sbml2biopax.sh alzpathway.xml
$ wget http://www.nature.com/msb/journal/v1/n1/extref/msb4100014-s3.xml -O Oda_EGFR_2005.xml
$ sh sbml2biopax.sh Oda_EGFR_2005.xml
$ wget http://www.nature.com/msb/journal/v1/n1/extref/msb4100057-s1.xml -O Oda_TLR_2006.xml
$ sh sbml2biopax.sh Oda_TLR_2006.xml

The sbml2biopax generate BioPAX Level 2 OWL file from given SBML file.

Store

4store

$ 4s-backend-setup biopax_lv2
4store[2577]: backend-setup.c:176 erased files for KB biopax_lv2
4store[2577]: backend-setup.c:301 created RDF metadata for KB biopax_lv2
$ 4s-backend biopax_lv2
$ 4s-import -v biopax_lv2 alzpathway.owl 
$ 4s-import -v biopax_lv2 Oda_EGFR_2005.owl 
removing old data
Reading <file:///Users/ogishima/.../Oda_EGFR_2005.owl>
Pass 1, processed 4951 triples (4951)
Pass 2, processed 4951 triples, 61233 triples/s
Updating index
Index update took 2.412844 seconds
Imported 4951 triples, average 1985 triples/s
$ 4s-import -v biopax_lv2 Oda_TLR_2006.owl
removing old data
Reading <file:///Users/ogishima/.../Oda_TLR_2006.owl>
Pass 1, processed 10087 triples (10087)
Pass 2, processed 10087 triples, 78545 triples/s
Updating index
Index update took 0.605142 seconds
Imported 10087 triples, average 13330 triples/s

Virtuoso

Coming soon.

Query

4store

$ 4s-query biopax_lv2
4store>SELECT * WHERE {?s ?p ?o}
   >#EOQ
?s	?p	?o
<http://www.ebi.ac.uk/biomodels/biopax.owl#c3>	<http://www.biopax.org/release/biopax-level2.owl#TERM>	"late_br_endsome"
<http://www.ebi.ac.uk/biomodels/biopax.owl#c2>	<http://www.biopax.org/release/biopax-level2.owl#TERM>	"recycling_br_endsome"
...
results.c:1655 hit soft limit 2 times
# hit complexity limit 2 times, increasing soft limit may give more results
#EOR

Virtuoso

Coming soon.

SPARQL end point

SPARQL end point is now under construction, and will appear soon. :-)

Visualization

Problems & ToDo?

  • BioPAX OWL might be too complicate to archive the above usecase (goal). SPARQL search might took so much time to return results. Appropriate reduction of information of SBML model is required to convert from SBML to RDF (not OWL).
  • In the case of conversion from SBML to RDF, mapping of name among models is required.

Members

Special thanks to Eric and Ono-san for valuable comments!