From ImageWeb

FlyWeb_project

Contents 1 SPARQL endpoint of BDGP Gene Expression Images 1.1 Setting up D2R server 1.2 The Schema of BDGP 1.3 Put BDGP and FlyTED in a Potluck

SPARQL endpoint of BDGP Gene Expression Images

See also:

• http://www.bioontology.org/wiki/index.php/OBD:SPARQL-GO
• http://www.bioontology.org/wiki/index.php/OBD:SPARQL-InSitu

Setting up D2R server

Please see DefiningImageAccess/Tool/D2RServer

The Schema of BDGP

When setting up the SPARQL endpoint, I have performed some customisation based on Chris Mungall's code (http://geneontology.cvs.sourceforge.net/geneontology/go-dev/sql/d2rq-mappings/). My mapping file is shared at our svn (https://milos2.zoo.ox.ac.uk/svn/ImageWeb/FlyWeb/Trunk/D2R-mapping-schemas/). It differs from Chris's mapping in two aspects:

1. It is more consistent with the BDGP dump used for setting up our endpoint;
2. It does not use a global URI pattern to express terms (http://purl.org/obo/xref/@@term.acc@@). Because URIs expressed in such format are sometimes broken. In order to maintain the data resource as a linked RDF graph, I chose to identify these terms using a local URI pattern: "term/@@term.acc@@".

The schema is shown in the following figure.

Figure 1. The outline schema of our local BDGP SPARQL Endpoint.

• It is very likely that resources about Gene_Products could overlap with those in our FlyTED repository.
• There might be some overlapping between annotations about Location in both BDGP and FlyTED, because they are also based upon the FlyAnatomy ontology (http://obofoundry.org/cgi-bin/detail.cgi?fly_anatomy). The anatomy ontology used to annotate our FlyTED images are under review by our biologist researchers and FlyBase researchers. Also, our annotations might use a completely different subsection of this ontology, and therefore they might contain no overlapping with BDGP's data.

Put BDGP and FlyTED in a Potluck

Potluck (http://simile.mit.edu/potluck/) is a tool that allows users to mashup Exhibit-powered RDF data resources. An introduction about Potluck can be found in its ISWC2007 paper, and online demo.

I am expecting Potluck to help us with integrating BDGP and FlyTED (without any code writing) in the following aspects:

• experiment with merging these two data resources by the common Gene names. The same gene might be identified differently in both resources. The different gene names referring to the same gene can be identified in different channel:
  • using FlyBase's flyid
  • using BDGP's alternative genenames for each record, which sometimes use the same gene name as in FlyTED.

• experiment with merging/aligning the anatomical terms used to annotate expressions in both resources.