rdfizers

RDFizers at the core of Semantic Web/Linked Data technologies

Use case: Generate semantically interoperable genome annotations in RDF using (semi-)automated approaches [1].

1. Introduction:

• RDF graph model (serializations) & Linked Data principles

• Genome annotations commonly store and shared via tab-delimited files in Generic Feature Format (GFF)

• Controlled vocabularies, taxonomies & ontologies (GenBank Feature Table Definition, NCBITaxon, SO[FA], FALDO)

• Metadata standards & PIDs (Dublin Core/DCMI, DCAT, Cool URIs or PURLs)

• RDFication approaches:

  • text file (e.g. GFF)-> RDB -> RDF (SIGA.py)
  • text file -> RDF (OpenRefine+RDF, Virtuoso Sponger)
  • RDB->RDF: Direct Mapping or R2RML (Virtuoso R2RML processor)

2. Install & test SIGA.py

git clone https://github.com/candYgene/siga.git
cd siga
virtualenv .sigaenv
source .sigaenv/bin/activate
pip install -r requirements.txt

• generate RDF graph (in turtle) from GFF file with tomato gene models

cd src
python SIGA.py -h
python SIGA.py db -rV ../examples/features.gff3 # GFF->RDB
ls ../examples/features.db # sqlite dbfile
python SIGA.py rdf -c config.ini -o turtle ../examples/features.db # RDB->RDF
ls ../examples/features.ttl

• investigate I/O files incl. mappings in config.ini

3. Install & test OpenRefine (OR) with RDF extension

docker pull fusepool/openrefine
docker run -p 3333:3333 -d fusepool/openrefine

• download an example OR project file on tomato quantitative trait loci (QTLs)

git clone https://github.com/candYgene/pbg-ld.git
cd pbg-ld/data && tar xvzf sgn-ld.tar.gz
cd other
ls tomato_QTLs.openrefine.tar.gz

• open http://localhost:3333/ in your favourite web browser

• import the project file: Import_Project->Project File->Import Project

• look at the tabulated data and the corresponding RDF data model: RDF button (upper-left corner)->Edit RDF skeleton

4. Install & test OpenLink Virtuoso Universal Server

docker pull candygene/docker-virtuoso
docker run -p 8890:8890 -d candygene/docker-virtuoso

• open the admin console (Conductor) in your web browser and login (dba for both account and password)

• upload RDF graphs: Linked_Data->Quad Store Upload->File and Named Graph URI

  • tomato gene models: siga/examples/features.ttl with http://solgenomics.net/genome/Solanum_lycopersicum graph URI
  • tomato QTLs: pbg-ld/data/rdf/tomato_QTLs.ttl.gz with http://europepmc.org/articles graph URI

• query RDF graphs via SPARQL endpoint

SELECT *
FROM <http://solgenomics.net/genome/Solanum_lycopersicum>
WHERE { ?s ?p ?o }

SELECT *
FROM <http://europepmc.org/articles>
WHERE { ?s ?p ?o }

References

[1] NLeSC candYgene project & code base on GitHub