med2rdf/hco

Human chromosome ontology

Human Chromosome Ontology

The Human Chromosome Ontology (HCO) provides simple and stable URIs for human reference genome versions to semantically identify human chromosomes.

Basic class hco:HumanChromosome is inherited by each chromosome class such as hco:21 and hco:GenomeBuild is inherited by each build version such as hco:GRCh38. Then, for example, the human chromosome 21 of the GRCh38 version is represented as hco:21\/GRCh38 (http://identifiers.org/hco/21/GRCh38) which is defined as an instance of hco:21 with build version hco:GRCh38. Each instance of a chromosome has links to external resources such as INSDC, RefSeq, UCSC etc. so that users can be navigated to the chromosome sequence and its annotations.

hco:21
        rdfs:label      "Human chromosome 21" ;
        rdfs:subClassOf hco:HumanChromosome .

hco:21\/GRCh38
        rdf:type        hco:21 ;
        hco:build       hco:GRCh38 ;
        hco:length      46709983 ;
        skos:altLabel   "GPC_000001313" ;
        hco:insdc       <http://identifiers.org/insdc/CM000683.2> ;
        hco:refseq      <http://identifiers.org/refseq/NC_000021.9> ;
        hco:ucsc        <http://genome.ucsc.edu/cgi-bin/hgTracks?db=hg38&position=chr21> ;
        hco:ensembl     <http://rdf.ebi.ac.uk/resource/ensembl/homo_sapiens/GRCh38/21> ,
                        <http://www.ensembl.org/Homo_sapiens/Location/Chromosome?r=21> .

Also, hco:Cytoband is inherited by each cytoband class such as hco:21q11.2 and an instance of it such as hco:21q11.2\/GRCh38 has the FALDO location on a chromosome and hco:bandtype such as hco:Gneg which stands for a nonstaining G-negative light band.

hco:21q11.2
        rdfs:label      "21q11.2" ;
        rdfs:subClassOf hco:Cytoband .

hco:21q11.2\/GRCh38
        rdf:type        hco:21q11.2 ;
        hco:build       hco:GRCh38 ;
        hco:bandtype    hco:Gneg ;
        faldo:location  [
                rdf:type         faldo:Region ;
                faldo:begin      [
                        rdf:type         faldo:BothStrandsPosition ;
                        faldo:position   13000000 ;
                        faldo:reference  hco:21\/GRCh38
                ] ;
                faldo:end        [
                        rdf:type         faldo:BothStrandsPosition ;
                        faldo:position   15000000 ;
                        faldo:reference  hco:21\/GRCh38
                ]
        ] .

Update procedure

To create the latest version of the ontology, first obtain the cytoBand data from UCSC and convert the data into RDF/Turtle.

$ curl http://hgdownload.cse.ucsc.edu/goldenPath/hg19/database/cytoBand.txt.gz | gzip -dc - > cytoBand-GRCh37.txt
$ curl http://hgdownload.cse.ucsc.edu/goldenPath/hg38/database/cytoBand.txt.gz | gzip -dc - > cytoBand-GRCh38.txt
$ ruby cytoBand2ttl.rb > cytoBand.ttl

Then, concatenate the header part of the ontology with the cytoBand data.

$ cat hco_head.ttl cytoBand.ttl > hco.tmp

Finallly, expand hco:X\/GRChXX into <http://identifiers.org/hco/X/GRChXX> for avoiding a triple store to treat it as <http://identifiers.org/hco/X\/GRChXX> (some triple store such as Virtuoso has an option to relax syntax rules for loading invalid data that sometimes causes further problem even for valid data).

$ rapper -i turtle -o turtle hco.tmp > hco.ttl

Contributors

• Toshiaki Katayama (Database Center for Life Science)
• The Med2RDF project http://med2rdf.org/
• The 3rd RDF summit https://github.com/dbcls/rdfsummit3

hco's Issues

Note patch usage in documentation

Careful effort has been put into describing the patch releases. Note how to use them properly in the README to close this.

Release status

What status would you give this? Can you add any issues that folks can help with to get it to a release status?

List at identifiers.org or ontology service

In the readme it links to an identifiers.org link that doesn't work. To close this, register the ontology so it can be resolved properly with some service that eases use.

hco.owl filename

The file in the repository hco.owl is actually a turtle file, better to be renamed!

VCF to RDF, hco in VCF

One of the frustrating aspects of working with VCF across genome builds is the lack of normalized ways of representing them and accessing them within VCF. How could hco be used to generate RDF from VCF with better chromosome metadata?

Could hco be used to provide more metadata to VCF headers?

Perhaps a good first tool would be to extend a variant annotator to include this in a VCF in a controlled way.

Document curation/generation process

If you could write a little more about the build process to ease maintenance it would be helpful. Some parts of the owl are hand edited, some system generated. If there are any other scripts or helpful documentation for how you got to hco.owl, please add them!