Skip to content

Commit cbc3991

Browse files
Merge pull request #275 from jctoledo/hgnc-r3
HGNC added new provenance model data
2 parents d255500 + 30a0cd3 commit cbc3991

File tree

1 file changed

+40
-12
lines changed

1 file changed

+40
-12
lines changed

hgnc/hgnc.php

Lines changed: 40 additions & 12 deletions
Original file line numberDiff line numberDiff line change
@@ -27,9 +27,11 @@
2727
* @version 2.0
2828
* @author Alison Callahan
2929
* @author Jose Cruz-Toledo
30+
* @author Michel Dumontier
3031
*/
3132

3233
require(__DIR__.'/../../php-lib/bio2rdfapi.php');
34+
require(__DIR__.'/../../php-lib/dataresource.php');
3335

3436
class HGNCParser extends Bio2RDFizer {
3537
private $version = 2.0;
@@ -58,6 +60,7 @@ function Run(){
5860
parent::setParameterValue('download',true);
5961
}
6062
//download the hgnc file
63+
$rfile = null;
6164
if(parent::getParameterValue('download') == true) {
6265
$rfile = $rdir;
6366
echo "downloading $file ... ";
@@ -81,20 +84,45 @@ function Run(){
8184
echo PHP_EOL;
8285
// generate the dataset release file
8386
echo "generating dataset release file... ";
84-
$desc = parent::getBio2RDFDatasetDescription(
85-
$this->getPrefix(),
86-
"https://github.com/bio2rdf/bio2rdf-scripts/blob/master/hgnc/hgnc.php",
87-
$this->getBio2RDFDownloadURL($this->getNamespace()),
88-
"http://www.genenames.org",
89-
array("use"),
90-
"http://www.genenames.org/about/overview",
91-
"ftp://ftp.ebi.ac.uk/pub/databases/genenames/hgnc_complete_set.txt.gz",
92-
parent::getDatasetVersion()
93-
);
94-
parent::setWriteFile($odir.$this->getBio2RDFReleaseFile($this->getNamespace()));
95-
parent::getWriteFile()->write($desc);
87+
$dataset_description = '';
88+
$source_file = (new DataResource($this))
89+
->setURI($rdir)
90+
->setTitle('HUGO Gene Nomenclature Committee (HGNC)')
91+
->setRetrievedDate(date("Y-m-d\TG:i:s\Z", filemtime($lfile)))
92+
->setFormat('text/tab-separated-value')
93+
->setFormat('application/zip')
94+
->setPublisher('http://www.genenames.org/')
95+
->setHomepage('http://www.genenames.org/data/gdlw_columndef.html')
96+
->setRights('use')
97+
->setRights('attribution')
98+
->setLicense('http://www.genenames.org/about/overview')
99+
->setDataset(parent::getDatasetURI());
100+
101+
$prefix = parent::getPrefix();
102+
$bVersion = parent::getParameterValue('bio2rdf_release');
103+
$date = date("Y-m-d\TG:i:s\Z");
104+
$output_file = (new DataResource($this))
105+
->setURI("http://download.bio2rdf.org/release/$bVersion/$prefix")
106+
->setTitle("Bio2RDF v$bVersion RDF version of $prefix (generated at $date)")
107+
->setSource($source_file->getURI())
108+
->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/hgnc/hgnc.php")
109+
->setCreateDate($date)
110+
->setHomepage("http://download.bio2rdf.org/release/$bVersion/$prefix/$prefix.html")
111+
->setPublisher("http://bio2rdf.org")
112+
->setRights("use-share-modify")
113+
->setRights("restricted-by-source-license")
114+
->setLicense("http://creativecommons/licenses/by/3.0/")
115+
->setDataset(parent::getDatasetURI());
116+
96117
parent::getWriteFile()->close();
97118
echo "done!".PHP_EOL;
119+
if($gz) $output_file->setFormat("application/gzip");
120+
if(strstr(parent::getParameterValue('output_format'),"nt")) $output_file->setFormat("application/n-triples");
121+
else $output_file->setFormat("application/n-quads");
122+
$dataset_description .= $source_file->toRDF().$output_file->toRDF();
123+
$this->setWriteFile($odir.$this->getBio2RDFReleaseFile());
124+
$this->getWriteFile()->write($dataset_description);
125+
$this->getWriteFile()->close();
98126

99127
}//Run
100128

0 commit comments

Comments
 (0)