Skip to content

Commit 0735cc2

Browse files
micheldumontiermicheldumontier
committed
Update omim.php
updated to https api url fixed id list processing getting full list of entries from mimtitles added some new exclusions
1 parent 865899b commit 0735cc2

File tree

1 file changed

+46
-69
lines changed

1 file changed

+46
-69
lines changed

‎omim/omim.php‎

Lines changed: 46 additions & 69 deletions
Original file line numberDiff line numberDiff line change
@@ -33,7 +33,7 @@ class OMIMParser extends Bio2RDFizer
3333
function __construct($argv) {
3434
parent::__construct($argv, 'omim');
3535
parent::addParameter('files',true,null,'all|omim#','entries to process: comma-separated list or hyphen-separated range');
36-
parent::addParameter('omim_api_url',false,null,'http://api.omim.org/api/entry?include=all&format=json');
36+
parent::addParameter('omim_api_url',false,null,'https://api.omim.org/api/entry?include=all&format=json');
3737
parent::addParameter('omim_api_key',false,null);
3838
parent::addParameter('omim_api_key_file',false,null,'omim.key','A file containing your omim KEY');
3939
parent::initialize();
@@ -55,22 +55,38 @@ function Run()
5555
}
5656
} else {
5757
trigger_error("No OMIM key has been provided either by commmand line or in the expected omim key file $key_file",E_USER_WARNING);
58+
exit;
5859
}
5960
}
6061

61-
// get the list of mim2gene entries
62-
$entries = $this->GetListOfEntries($ldir);
62+
// get the list of entries
63+
$file = "mimTitles.txt";
64+
$rfile = "https://data.omim.org/downloads/$key/$file";
65+
$lfile = $ldir.$file;
66+
if(!file_exists($lfile) && parent::getParameterValue('download') == false) {
67+
trigger_error($lfile." not found. Will attempt to download.", E_USER_NOTICE);
68+
parent::setParameterValue('download',true);
69+
}
70+
if(parent::getParameterValue('download') == true) {
71+
echo "downloading $file ... ";
72+
Utils::DownloadSingle($rfile, $lfile);
73+
}
74+
// parse the file
75+
$fp = fopen($lfile,"rb");
76+
while($l = fgetcsv($fp,0,"\t")) {
77+
if($l[0][0] == "#") continue;
78+
$full_list[ $l[1] ] = "";
79+
}
6380

81+
6482
// get the work specified
6583
$list = trim(parent::getParameterValue('files'));
6684
if($list != 'all') {
6785
// check if a hyphenated list was provided
6886
if(($pos = strpos($list,"-")) !== FALSE) {
6987
$start_range = substr($list,0,$pos);
7088
$end_range = substr($list,$pos+1);
71-
72-
// get the whole list
73-
$full_list = $this->GetListOfEntries($ldir);
89+
7490
// now intersect
7591
foreach($full_list AS $e => $type) {
7692
if($e >= $start_range && $e <= $end_range) {
@@ -84,9 +100,11 @@ function Run()
84100
foreach($b AS $e) {
85101
$myentries[$e] = '';
86102
}
87-
$entries = array_intersect_key ($entries,$myentries);
103+
$entries = array_intersect_key ($full_list,$myentries);
88104
}
89-
}
105+
} else $entries = $full_list;
106+
107+
echo "Will process a total of ".count($entries)." OMIM entries".PHP_EOL;
90108

91109
// set the write file
92110
$gz = (strstr(parent::getParameterValue('output_format'),".gz") === FALSE)?false:true;
@@ -102,17 +120,13 @@ function Run()
102120
$total = count($entries);
103121
foreach($entries AS $omim_id => $type) {
104122
echo "processing ".(++$i)." of $total - omim# ";
105-
$download_file = $ldir.$omim_id.".json.gz";
106-
$gzfile = "compress.zlib://$download_file";
123+
$lfile = $ldir.$omim_id.".json.gz";
124+
$gzfile = "compress.zlib://$lfile";
107125
// download if the file doesn't exist or we are told to
108-
if(!file_exists($download_file) || parent::getParameterValue('download') == true) {
126+
if(!file_exists($lfile) || parent::getParameterValue('download') == true) {
109127
// download using the api
110-
$url = parent::getParameterValue('omim_api_url').'&apiKey='.parent::getParameterValue('omim_api_key').'&mimNumber='.$omim_id;
111-
$buf = file_get_contents($url);
112-
if(strlen($buf) != 0) {
113-
file_put_contents($download_file, $buf);
114-
usleep(500000); // limit of 4 requests per second
115-
}
128+
$rfile = parent::getParameterValue('omim_api_url').'&apiKey='.parent::getParameterValue('omim_api_key').'&mimNumber='.$omim_id;
129+
Utils::DownloadSingle($rfile, $lfile);
116130
}
117131

118132
// load entry, parse and write to file
@@ -170,57 +184,6 @@ function Run()
170184
return TRUE;
171185
}
172186

173-
function getListOfEntries($ldir)
174-
{
175-
// get the master list of entries
176-
$file = "mim2gene.txt";
177-
if(!file_exists($ldir.$file)) {
178-
trigger_error($ldir.$file." not found. Will attempt to download. ", E_USER_NOTICE);
179-
$this->SetParameterValue('download',true);
180-
}
181-
182-
if(parent::getParameterValue('download')==true) {
183-
// connect
184-
if(!isset($ftp)) {
185-
$host = 'ftp.omim.org';
186-
echo "connecting to $host ...";
187-
$ftp = ftp_connect($host);
188-
if(!$ftp) {
189-
echo "Unable to connect to $host".PHP_EOL;
190-
die;
191-
}
192-
ftp_pasv ($ftp, true) ;
193-
$login = ftp_login($ftp, 'anonymous', '[email protected]');
194-
if ((!$ftp) || (!$login)) {
195-
echo "FTP-connect failed!"; die;
196-
} else {
197-
echo "Connected".PHP_EOL;
198-
}
199-
}
200-
201-
// download
202-
ftp_pasv($ftp, true);
203-
echo "Downloading $file ...";
204-
if(ftp_get($ftp, $ldir.$file, 'OMIM/'.$file, FTP_BINARY) === FALSE) {
205-
trigger_error("Error in downloading $file");
206-
}
207-
if(isset($ftp)) ftp_close($ftp);
208-
echo "success!".PHP_EOL;
209-
}
210-
211-
// parse the mim2gene file for the entries
212-
// # Mim Number Type Gene IDs Approved Gene Symbols
213-
$fp = fopen($ldir.$file,"r");
214-
fgets($fp);
215-
while($l = fgets($fp)) {
216-
$a = explode("\t",$l);
217-
if($a[1] != "moved/removed")
218-
$list[$a[0]] = $a[1];
219-
}
220-
fclose($fp);
221-
return $list;
222-
}
223-
224187

225188
function get_phenotype_mapping_method_type($id = null, $generate_declaration = false)
226189
{
@@ -326,6 +289,14 @@ function ParseEntry($obj, $type)
326289
}
327290
}
328291

292+
// check if moved
293+
if(isset($o['movedTo'])) {
294+
$new_omim_uri = parent::getNamespace().$o['movedTo'];
295+
parent::addRDF(
296+
parent::triplify($omim_uri, parent::getVoc()."superceded-by", $new_omim_uri)
297+
);
298+
}
299+
329300
// parse text sections
330301
if(isset($o['textSectionList'])) {
331302
foreach($o['textSectionList'] AS $i => $section) {
@@ -542,7 +513,8 @@ function ParseEntry($obj, $type)
542513

543514
$ns = '';
544515
switch($k) {
545-
case 'approvedGeneSymbols': $ns = 'symbol';break;
516+
case 'hgncID': $ns = 'hgnc';break;
517+
case 'approvedGeneSymbols': $ns = 'hgnc.symbol';break;
546518
case 'geneIDs': $ns = 'ncbigene';break;
547519
case 'ncbiReferenceSequences': $ns = 'gi';break;
548520
case 'genbankNucleotideSequences': $ns = 'gi';break;
@@ -568,6 +540,11 @@ function ParseEntry($obj, $type)
568540
case 'diseaseOntologyIDs': $ns = 'do';break;
569541

570542
// specifically ignorning
543+
case 'newbornScreening':
544+
case 'clinGenDosage':
545+
case 'clinGenValidity':
546+
case 'monarch':
547+
case 'decipherSyndromes':
571548
case 'geneTests':
572549
case 'cmgGene':
573550
case 'geneticAllianceIDs': // #

0 commit comments

Comments
 (0)