27
27
* @version 2.0
28
28
* @author Alison Callahan
29
29
* @author Jose Cruz-Toledo
30
+ * @author Michel Dumontier
30
31
*/
31
32
32
- require (__DIR__ .'/../../php-lib/bio2rdfapi.php ' );
33
-
34
33
class HGNCParser extends Bio2RDFizer {
35
34
private $ version = 2.0 ;
36
35
function __construct ($ argv ){
@@ -58,6 +57,7 @@ function Run(){
58
57
parent ::setParameterValue ('download ' ,true );
59
58
}
60
59
//download the hgnc file
60
+ $ rfile = null ;
61
61
if (parent ::getParameterValue ('download ' ) == true ) {
62
62
$ rfile = $ rdir ;
63
63
echo "downloading $ file ... " ;
@@ -81,20 +81,45 @@ function Run(){
81
81
echo PHP_EOL ;
82
82
// generate the dataset release file
83
83
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 );
84
+ $ dataset_description = '' ;
85
+ $ source_file = (new DataResource ($ this ))
86
+ ->setURI ($ rdir )
87
+ ->setTitle ('HUGO Gene Nomenclature Committee (HGNC) ' )
88
+ ->setRetrievedDate (date ("Y-m-d\TG:i:s\Z " , filemtime ($ lfile )))
89
+ ->setFormat ('text/tab-separated-value ' )
90
+ ->setFormat ('application/zip ' )
91
+ ->setPublisher ('http://www.genenames.org/ ' )
92
+ ->setHomepage ('http://www.genenames.org/data/gdlw_columndef.html ' )
93
+ ->setRights ('use ' )
94
+ ->setRights ('attribution ' )
95
+ ->setLicense ('http://www.genenames.org/about/overview ' )
96
+ ->setDataset (parent ::getDatasetURI ());
97
+
98
+ $ prefix = parent ::getPrefix ();
99
+ $ bVersion = parent ::getParameterValue ('bio2rdf_release ' );
100
+ $ date = date ("Y-m-d\TG:i:s\Z " );
101
+ $ output_file = (new DataResource ($ this ))
102
+ ->setURI ("http://download.bio2rdf.org/release/ $ bVersion/ $ prefix " )
103
+ ->setTitle ("Bio2RDF v $ bVersion RDF version of $ prefix (generated at $ date) " )
104
+ ->setSource ($ source_file ->getURI ())
105
+ ->setCreator ("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/hgnc/hgnc.php " )
106
+ ->setCreateDate ($ date )
107
+ ->setHomepage ("http://download.bio2rdf.org/release/ $ bVersion/ $ prefix/ $ prefix.html " )
108
+ ->setPublisher ("http://bio2rdf.org " )
109
+ ->setRights ("use-share-modify " )
110
+ ->setRights ("restricted-by-source-license " )
111
+ ->setLicense ("http://creativecommons/licenses/by/3.0/ " )
112
+ ->setDataset (parent ::getDatasetURI ());
113
+
96
114
parent ::getWriteFile ()->close ();
97
115
echo "done! " .PHP_EOL ;
116
+ if ($ gz ) $ output_file ->setFormat ("application/gzip " );
117
+ if (strstr (parent ::getParameterValue ('output_format ' ),"nt " )) $ output_file ->setFormat ("application/n-triples " );
118
+ else $ output_file ->setFormat ("application/n-quads " );
119
+ $ dataset_description .= $ source_file ->toRDF ().$ output_file ->toRDF ();
120
+ $ this ->setWriteFile ($ odir .$ this ->getBio2RDFReleaseFile ());
121
+ $ this ->getWriteFile ()->write ($ dataset_description );
122
+ $ this ->getWriteFile ()->close ();
98
123
99
124
}//Run
100
125
0 commit comments