function Run()
{
// directory shortcuts
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
// get the listings page
$rfile = trim(parent::getParameterValue('download_url'));
$file = "interpro.xml.gz";
$lfile = $ldir . $file;
if (!file_exists($lfile) || parent::getParameterValue("download") == "true") {
echo "Downloading {$lfile}" . PHP_EOL;
$ret = file_get_contents($rfile);
if ($ret === FALSE) {
trigger_error("unable to download {$rfile}");
exit;
}
file_put_contents($lfile, $ret);
}
echo "Loading XML file...";
$cxml = new CXML($lfile);
$cxml->Parse();
$xml = $cxml->GetXMLRoot();
echo "Done" . PHP_EOL;
// set the write file
$gz = strstr(parent::getParameterValue('output_format'), ".gz") === FALSE ? false : true;
$outfile = "interpro." . parent::getParameterValue('output_format');
parent::setWriteFile($odir . $outfile, $gz);
echo "Parsing interpro xml file" . PHP_EOL;
$this->parse($xml);
parent::writeRDFBufferToWriteFile();
parent::getWriteFile()->close();
echo "Done!" . PHP_EOL;
// let's make an nq file
parent::setGraphURI(parent::getDatasetURI());
// dataset description
$source_version = parent::getDatasetVersion();
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("InterPro v{$source_version}")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat("application/xml")->setFormat("application/g-zip")->setPublisher("http://www.ebi.ac.uk/")->setHomepage("http://www.ebi.ac.uk/interpro/")->setRights("InterPro - Integrated Resource Of Protein Domains And Functional Sites. Copyright (C) 2001 The InterPro Consortium")->setLicense("http://www.ebi.ac.uk/interpro/faqs.html")->setDataset("http://identifiers.org/interpro/");
$prefix = parent::getPrefix();
$bVersion = parent::getParameterValue('bio2rdf_release');
$date = date("Y-m-d\\TG:i:s\\Z");
$output_file = (new DataResource($this))->setURI("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$outfile}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} v{$source_version}")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/interpro/interpro.php")->setCreateDate($date)->setHomepage("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$prefix}.html")->setPublisher("http://bio2rdf.org")->setRights("use-share-modify")->setRights("by-attribution")->setRights("restricted-by-source-license")->setLicense("http://creativecommons.org/licenses/by/3.0/")->setDataset(parent::getDatasetURI());
if ($gz) {
$output_file->setFormat("application/gzip");
}
if (strstr(parent::getParameterValue('output_format'), "nt")) {
$output_file->setFormat("application/n-triples");
} else {
$output_file->setFormat("application/n-quads");
}
$dataset_description = $source_file->toRDF() . $output_file->toRDF();
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
return true;
}
function Run()
{
// directory shortcuts
$ldir = $this->GetParameterValue('indir');
$odir = $this->GetParameterValue('outdir');
// get the listings page
$rfile = trim($this->GetParameterValue('download_url'));
$file = "interpro.xml.gz";
$lfile = $ldir . $file;
if (!file_exists($lfile) || $this->GetParameterValue("download") == "true") {
echo "Downloading {$lfile}" . PHP_EOL;
$ret = file_get_contents($rfile);
if ($ret === FALSE) {
trigger_error("unable to download {$rfile}");
exit;
}
file_put_contents($lfile, $ret);
}
$cxml = new CXML($ldir, $file);
$cxml->Parse();
$xml = $cxml->GetXMLRoot();
// set the write file
$outfile = 'interpro.nt';
$gz = false;
if ($this->GetParameterValue('graph_uri')) {
$outfile = 'interpro.nq';
}
if ($this->GetParameterValue('gzip')) {
$outfile .= '.gz';
$gz = true;
}
$this->SetWriteFile($odir . $outfile, $gz);
echo "Parsing interpro xml file" . PHP_EOL;
$this->Parse($xml);
$this->WriteRDFBufferToWriteFile();
$this->GetWriteFile()->Close();
echo "Done!" . PHP_EOL;
// generate the release file
$this->DeleteBio2RDFReleaseFiles($odir);
$desc = $this->GetBio2RDFDatasetDescription($this->GetNamespace(), "https://github.com/bio2rdf/bio2rdf-scripts/blob/master/interpro/intepro.php", $this->GetBio2RDFDownloadURL($this->GetNamespace()) . $outfile, "http://www.ebi.ac.uk/interpro/", array("use-share-modify"), null, $this->GetParameterValue('download_url'), $this->version);
$this->SetWriteFile($odir . $this->GetBio2RDFReleaseFile($this->GetNamespace()));
$this->GetWriteFile()->Write($desc);
$this->GetWriteFile()->Close();
return true;
}
/**
* process a results xml file from the download directory
**/
function process_file($infile)
{
$indir = parent::getParameterValue('indir');
$xml = new CXML($infile);
$this->setCheckPoint('file');
while ($xml->Parse("clinical_study") == TRUE) {
$this->setCheckPoint('record');
$this->root = $root = $xml->GetXMLRoot();
$this->nct_id = $nct_id = $this->getString("//id_info/nct_id");
$this->study_id = $study_id = parent::getNamespace() . "{$nct_id}";
### declare
$label = $this->getString("//brief_title");
if (!$label) {
$label = $this->getString("//official_title");
}
if (!$label) {
$label = "Clinical trial #" . $nct_id;
}
parent::addRDF(parent::describeIndividual($study_id, $label, parent::getVoc() . "Clinical-Study") . parent::describeClass(parent::getVoc() . "Clinical-Study", "Clinical Study"));
##########################################################################################
#required header
##########################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "download-date", $this->getString('//required_header/download_date')) . parent::triplify($study_id, parent::getVoc() . "url", $this->getString('//required_header/url')));
##########################################################################################
#identifiers
##########################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "nct-id", $this->getString('//id_info/nct_id'), "xsd:string") . parent::triplifyString($study_id, parent::getVoc() . "org-study-id", $this->getString('//id_info/org_study_id'), "xsd:string"));
$sids = $root->xpath('//id_info/secondary_id');
if (isset($sids)) {
foreach ($sids as $id) {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "secondary-id", (string) $id, "xsd:string"));
}
}
$nctaliases = $root->xpath('//id_info/nct-alias');
if (isset($nctaliases)) {
foreach ($nctaliases as $id) {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "nct-alias", (string) $id, "xsd:string"));
}
}
##########################################################################################
#titles
##########################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "brief-title", $this->getString("//brief_title")) . parent::triplifyString($study_id, parent::getVoc() . "official-title", $this->getString("//official_title")));
###################################################################################
#brief summary
###################################################################################
$brief_summary = str_replace(array("\r", "\n", "\t"), array("
", "
", "	"), $this->getString('//brief_summary/textblock'));
parent::addRDF(parent::triplifyString($study_id, $this->getVoc() . "brief-summary", $brief_summary));
####################################################################################
# detailed description
####################################################################################
$d = str_replace(array("\r", "\n", "\t"), array("
", "
", "	"), $this->getString('//detailed_description/textblock'));
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "detailed-description", $d));
#########################################################################################
#acronym
#########################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "acronym", $this->getString("//acronym")));
########################################################################################
#sponsors
########################################################################################
try {
$sponsors = array("lead_sponsor", "collaborator");
foreach ($sponsors as $sponsor) {
$a = @array_shift($root->xpath('//sponsors/' . $sponsor));
if ($a == null) {
break;
}
$agency = $this->getString("//agency", $a);
$agency_id = parent::getRes() . md5($agency);
$agency_class = $this->getString("//agency_class", $a);
$agency_class_id = parent::getRes() . md5($agency_class);
parent::addRDF(parent::describeIndividual($agency_id, $agency, parent::getVoc() . "Organization") . parent::describeClass(parent::getVoc() . "Organization", "Organization") . parent::triplify($study_id, parent::getVoc() . str_replace("_", "-", $sponsor), $agency_id) . parent::describeIndividual($agency_class_id, $agency_class, parent::getVoc() . "Organization") . parent::describeClass(parent::getVoc() . "Organization", "Organization") . parent::triplify($agency_id, parent::getVoc() . "organization", $agency_class_id));
}
} catch (Exception $e) {
echo "There was an error in the lead sponsor element: {$e}\n";
}
#################################################################################
# source
#################################################################################
$source = $this->getString('//source');
if ($source) {
$source_id = parent::getRes() . md5($source);
parent::addRDF(parent::describeIndividual($source_id, $source, parent::getVoc() . "Organization") . parent::triplify($study_id, parent::getVoc() . "source", $source_id));
}
######################################################################################
# oversight
######################################################################################
try {
$oversight = @array_shift($root->xpath('//oversight_info'));
$oversight_id = parent::getRes() . md5($oversight->asXML());
$authority = $this->getString('//authority', $oversight);
$authority_id = parent::getRes() . md5($authority);
parent::addRDF(parent::describeIndividual($oversight_id, $authority, parent::getVoc() . "Organization") . parent::triplify($study_id, $this->getVoc() . "oversight", $oversight_id) . parent::triplify($study_id, $this->getVoc() . "authority", $authority_id) . parent::triplifyString($oversight_id, parent::getVoc() . "has-dmc", $this->getString('//has_dmc', $oversight)));
} catch (Exception $e) {
echo "There was an error in the oversight info element: {$e}\n";
}
#################################################################################
# overall status
#################################################################################
$overall_status = $this->getString('//overall_status');
if ($overall_status) {
$status_id = parent::getRes() . md5($overall_status);
parent::addRDF(parent::describeIndividual($status_id, $overall_status, parent::getVoc() . "Status") . parent::describeClass(parent::getVoc() . "Status", "Status") . parent::triplify($study_id, parent::getVoc() . "overall-status", $status_id));
}
#########################################################################################
#why stopped
#########################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "why-stopped", $this->getString("//why_stopped")));
##################################################################################
# dates
##################################################################################
$dates = array("start_date", "end_date", "completion_date", "primary_completion_date", "verification_date", "lastchanged_date", "firstreceived_date", "firstreceived_results_date");
foreach ($dates as $date) {
$d = $this->getString('//' . $date);
if ($d) {
$datetime = $this->getDatetimeFromDate($d);
if (isset($datetime)) {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . str_replace("_", "-", $date), $datetime));
} else {
trigger_error("unable to parse date: {$d}", E_USER_ERROR);
}
}
}
####################################################################################
# phase
####################################################################################
$phase = $this->getString('//phase');
if ($phase && $phase != "N/A") {
$phase_id = $this->getRes() . md5($phase);
parent::addRDF(parent::describeIndividual($phase_id, $phase, parent::getVoc() . "Phase", $phase) . parent::describeClass(parent::getVoc() . "Phase", $phase) . parent::triplify($study_id, parent::getVoc() . "phase", $phase_id));
}
###################################################################################
# study type
####################################################################################
$study_type = $this->getString('//study_type');
if ($study_type) {
$study_type_id = $this->getRes() . md5($study_type);
parent::addRDF(parent::describeClass($study_type_id, $study_type, parent::getVoc() . "Study-Type") . parent::describeClass(parent::getVoc() . "Study-Type", "Study Type") . parent::triplify($study_id, parent::getVoc() . "study-type", $study_type_id));
}
###############################################################################
# study design
###############################################################################
$study_design = $this->getString('//study_design');
if ($study_design) {
$study_design_id = parent::getRes() . md5($study_id . $study_design);
parent::addRDF(parent::describeIndividual($study_design_id, "{$study_id} study design", parent::getVoc() . "Study-Design") . parent::describeClass(parent::getVoc() . "Study-Design", "Study Design") . parent::triplify($study_id, parent::getVoc() . "study-design", $study_design_id));
// Intervention Model: Parallel Assignment, Masking: Double-Blind, Primary Purpose: Treatment
foreach (explode(", ", $study_design) as $i => $b) {
$c = explode(": ", $b);
if (isset($c[1])) {
$sdp = $study_design_id . "-" . ($i + 1);
$key = parent::getRes() . md5($c[0]);
$value = parent::getRes() . md5($c[1]);
parent::addRDF(parent::describeIndividual($sdp, $b, parent::getVoc() . "Study-Design-Parameter") . parent::describeClass(parent::getVoc() . "Study-Design-Parameter", "Study Design Parameter") . parent::triplify($sdp, parent::getVoc() . "key", $key) . parent::describeClass($key, $c[0]) . parent::triplify($sdp, parent::getVoc() . "value", $value) . parent::describeClass($value, $c[1]) . parent::triplify($study_design_id, parent::getVoc() . "study-design-parameter", $sdp));
}
}
}
####################################################################################
# target duration
####################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "target-duration", $this->getString('//target_duration')));
################################################################################
# outcomes
###############################################################################
$outcomes = array("primary_outcome", "secondary_outcome", "other_outcome");
foreach ($outcomes as $outcome) {
$o = $root->xpath('//' . $outcome);
if ($o) {
$os = $o;
if (!is_array($o)) {
$os = array($o);
}
foreach ($os as $o) {
try {
$po_id = parent::getRes() . md5($nct_id . $o->asXML());
$po_type = parent::getVoc() . str_replace("_", "-", $outcome);
$measure = $this->getString('//measure', $o);
$time_frame = $this->getString('//time_frame', $o);
$safety_issue = $this->getString('//saftey_issue', $o);
$description = $this->getString('//description', $o);
parent::addRDF(parent::describeIndividual($po_id, $measure . " " . $time_frame, ucfirst($po_type)) . parent::describeClass(ucfirst($po_type), str_replace("_", " ", ucfirst($outcome))) . parent::triplifyString($po_id, "dc:description", $description) . parent::triplifyString($po_id, parent::getVoc() . "measure", $measure) . parent::triplifyString($po_id, parent::getVoc() . "time-frame", $time_frame) . parent::triplifyString($po_id, parent::getVoc() . "safety-issue", $safety_issue) . parent::triplify($study_id, parent::getVoc() . $po_type, $po_id));
} catch (Exception $e) {
echo "There was an error parsing the primary outcome element: {$e} \n";
}
}
}
}
##############################################################################
#number of arms
##############################################################################
try {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "number-of-arms", $this->getString('//number_of_arms')));
} catch (Exception $e) {
echo "There was an exception parsing the number of arms element: {$e}\n";
}
##############################################################################
#number of groups
##############################################################################
try {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "number-of-arms", $this->getString('//number_of_groups')));
} catch (Exception $e) {
echo "There was an exception parsing the number of groups: {$e}\n";
}
##############################################################################
#enrollment
##############################################################################
try {
$e = $root->xpath('//enrollment');
if ($e) {
$type = strtolower((string) $e[0]->attributes()->type);
$value = $this->getString('//enrollment');
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . ($type ? $type . "-" : "") . "enrollment", $value));
}
} catch (Exception $e) {
echo "There was an exception parsing the enrollment element: {$e}\n";
}
###############################################################################
#condition
###############################################################################
try {
$conditions = $root->xpath('//condition');
foreach ($conditions as $condition) {
$mesh_label_id = parent::getRes() . md5($condition);
parent::addRDF(parent::triplify($study_id, parent::getVoc() . "condition", $mesh_label_id) . parent::describeClass($mesh_label_id, $condition, parent::getVoc() . "Condition") . parent::describeClass(parent::getVoc() . "Condition", "Condition"));
}
} catch (Exception $e) {
echo "There was an exception parsing condition element: {$e}\n";
}
################################################################################
# arm_group
################################################################################
try {
$arm_groups = $root->xpath('//arm_group');
foreach ($arm_groups as $arm_group) {
$arm_group_id = $this->getString('./arm_group_label', $arm_group);
$arm_group_id = md5($arm_group_id);
$arm_group_uri = parent::getRes() . $this->nct_id . "/arm-group/" . $arm_group_id;
$arm_group_label = $this->nct_id . " arm group " . $arm_group_id;
$arm_group_type = ucfirst(str_replace(" ", "_", $this->getString('./arm_group_type', $arm_group)));
if (!$arm_group_type) {
$arm_group_type = "Clinical-Arm";
}
$description = $this->getString('./description', $arm_group);
parent::addRDF(parent::describeIndividual($arm_group_uri, $arm_group_label, parent::getVoc() . $arm_group_type) . parent::describeClass(parent::getVoc() . $arm_group_type, ucfirst(str_replace("_", " ", $arm_group_type))) . parent::triplifyString($arm_group_uri, parent::getVoc() . "description", $description) . parent::describeIndividual($arm_group_uri, $arm_group, parent::getVoc() . "Arm-Group") . parent::describeClass(parent::getVoc() . "Arm-Group", "Arm Group") . parent::triplify($study_id, parent::getVoc() . "arm-group", $arm_group_uri));
}
} catch (Exception $e) {
echo "There was an exception in arm groups: {$e}\n";
}
##############################################################################
#intervention
##############################################################################
try {
$interventions = $root->xpath('//intervention');
foreach ($interventions as $intervention) {
$intervention_id = parent::getRes() . md5($intervention->asXML());
$intervention_name = $this->getString('./intervention_name', $intervention);
$intervention_type = $this->getString('./intervention_type', $intervention);
$intervention_type_uri = parent::getVoc() . ucfirst(str_replace(" ", "_", $intervention_type));
$intervention_desc = $this->getString('./description', $intervention);
$intervention_on = $this->getString('./other_name', $intervention);
parent::addRDF(parent::describeIndividual($intervention_id, $intervention_name, $intervention_type_uri) . parent::describeClass($intervention_type_uri, $intervention_type) . parent::triplifyString($intervention_id, parent::getVoc() . "intervention-name", $intervention_name) . parent::triplifyString($intervention_id, parent::getVoc() . "intervention-desc", $intervention_desc) . parent::triplifyString($intervention_id, parent::getVoc() . "other-name", $intervention_on) . parent::triplify($study_id, parent::getvoc() . "intervention", $intervention_id));
$agl = $intervention->xpath("./arm_group_label");
foreach ($agl as $a) {
$arm_group_id = md5($a);
$ag = parent::getRes() . $this->nct_id . "/arm-group/" . $arm_group_id;
parent::addRDF(parent::describeIndividual($ag, $a, parent::getVoc() . "Arm-Group") . parent::describeClass(parent::getVoc() . "Arm-Group", "Arm Group") . parent::triplify($intervention_id, parent::getVoc() . "arm-group", $ag));
}
}
} catch (Exception $e) {
echo "There was an error in interventions {$e}\n";
}
###############################################################################
#eligibility
################################################################################
try {
$eligibility = @array_shift($root->xpath('//eligibility'));
if ($eligibility !== null) {
$eligibility_label = "eligibility for " . $study_id;
$eligibility_id = parent::getRes() . md5($eligibility->asXML());
parent::addRDF(parent::describeIndividual($eligibility_id, $eligibility_label, parent::getVoc() . "Eligibility") . parent::describeClass(parent::getVoc() . "Eligibility", "Eligibility") . parent::triplify($study_id, parent::getVoc() . "eligibility", $eligibility_id));
if ($criteria = @array_shift($eligibility->xpath('./criteria'))) {
$text = @array_shift($criteria->xpath('./textblock'));
parent::addRDF(parent::triplifyString($eligibility_id, parent::getVoc() . "text", $text));
$c = preg_split("/(Inclusion Criteria\\:|Exclusion Criteria\\:)/", $text);
//inclusion
if (isset($c[1])) {
$d = explode(" - ", $c[1]);
// the lists are separated by a hyphen
foreach ($d as $inclusion) {
$inc = trim($inclusion);
if ($inc != '') {
$inc_id = parent::getRes() . md5($inc);
parent::addRDF(parent::describeIndividual($inc_id, $inc, parent::getVoc() . "Inclusion-Criteria") . parent::describeClass(parent::getVoc() . "Inclusion-Criteria", "Inclusion Criteria") . parent::triplify($eligibility_id, parent::getVoc() . "inclusion-criteria", $inc_id));
}
}
}
//exclusion
if (isset($c[2])) {
$d = explode(" - ", $c[1]);
foreach ($d as $exclusion) {
$exc = trim($exclusion);
if ($exc != '') {
$exc_id = parent::getRes() . md5($exc);
parent::addRDF(parent::describeIndividual($exc_id, $exc, parent::getVoc() . "Exclusion-Criteria") . parent::describeClass(parent::getVoc() . "Exclusion-Criteria", "Exclusion Criteria") . parent::triplify($eligibility_id, parent::getVoc() . "exclusion-criteria", $exc_id));
}
}
}
}
parent::addRDF(parent::triplifyString($eligibility_id, parent::getVoc() . "gender", $this->getString('./gender', $eligibility)));
parent::addRDF(parent::triplifyString($eligibility_id, parent::getVoc() . "healthy-volunteers", $this->getString('./healthy_volunteers', $eligibility)));
$attributes = array('minimum_age', 'maximum_age');
foreach ($attributes as $a) {
$s = $this->getString('./' . $a, $eligibility);
if ($s != 'N/A') {
$age = trim(str_replace("Years", "", $s));
parent::addRDF(parent::triplifyString($eligibility_id, parent::getVoc() . str_replace("_", "-", $a), $age));
}
}
$attributes = array("study_pop" => "study-population", "sampling_method" => "sampling-method");
foreach ($attributes as $a => $r) {
$e = @array_shift($eligibility->xpath('./' . $a));
if ($s = $this->getString('./' . $a, $eligibility)) {
parent::addRDF(parent::triplifyString($eligibility_id, parent::getVoc() . $r, $this->getString('./textblock', $e)));
}
}
}
} catch (Exception $e) {
echo "There was an error in eligibility: {$e}\n";
}
######################################################################################
#biospec
#####################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "biospec-retention", $this->getString('//biospec_retention')));
try {
$b = @array_shift($root->xpath('//biospec_descr'));
if ($b) {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "biospec_descr", $this->getString('./textblock', $b)));
}
} catch (Exception $e) {
echo "There was an error in biospec_descr: {$e}\n";
}
###################################################################
# contacts
###################################################################
$contacts = array("overall_official", "overall_contact", "overall_contact_backup");
try {
foreach ($contacts as $c) {
$d = @array_shift($root->xpath('//' . $c));
if ($d) {
parent::addRDF(parent::triplify($study_id, parent::getVoc() . str_replace("_", "-", $c), $this->makeContact($d)));
}
}
} catch (Exception $e) {
echo "There was an error parsing overall contact: {$e}" . "\n";
}
##############################################################
# location of facility doing the testing
##############################################################
try {
$location = @array_shift($root->xpath('//location'));
if ($location) {
$location_uri = parent::getRes() . md5($location->asXML());
$name = $this->getString('//facility/name', $location);
$address = @array_shift($location->xpath('//facility/address'));
$contact = @array_shift($location->xpath('//contact'));
$backups = @array_shift($location->xpath('//contact_backup'));
$investigators = @array_shift($location->xpath('//investigator'));
parent::addRDF(parent::describeIndividual($location_uri, $name, parent::getVoc() . "Location") . parent::describeClass(parent::getVoc() . "Location", "Location") . parent::triplifyString($location_uri, parent::getVoc() . "status", $this->getString('//status', $location)) . parent::triplify($study_id, parent::getVoc() . "location", $location_uri) . parent::triplify($location_uri, parent::getVoc() . "address", $this->makeAddress($address)) . ($contact != null ? parent::triplify($location_uri, parent::getVoc() . "contact", $this->makeContact($contact)) : ""));
if ($backups) {
foreach ($backups as $backup) {
parent::addRDF(parent::triplify($location_uri, parent::getVoc() . "contact-backup", $this->makeContact($backup)));
}
}
if ($investigators) {
foreach ($investigators as $investigator) {
parent::addRDF(parent::triplify($location_uri, parent::getVoc() . "investigator", $this->makeContact($investigator)));
}
}
}
} catch (Exception $e) {
echo "There was an error parsing location: {$e}" . "\n";
}
######################################################################
#countries
######################################################################
try {
$a = array("location_countries", "removed_countries");
foreach ($a as $country) {
$lc = @array_shift($root->xpath('//' . $country));
if ($lc) {
$label = $this->getString('//country', $lc);
$cid = parent::getRes() . md5($label);
parent::addRDF(parent::describeIndividual($cid, $label, parent::getVoc() . "Country") . parent::describeClass(parent::getVoc() . "Country", "Country") . parent::triplify($study_id, parent::getVoc() . "country", $cid));
}
}
} catch (Exception $e) {
echo "There was an error parsing country: {$e}" . "\n";
}
######################################################################
#reference
######################################################################
try {
$a = array("reference", "result_reference");
foreach ($a as $ref_type) {
$references = $root->xpath('//' . $ref_type);
foreach ($references as $reference) {
$p = $this->getString('./PMID', $reference);
if ($p) {
$pmid = "pubmed:{$p}";
parent::addRDF(parent::describeIndividual($pmid, $p, parent::getVoc() . "Reference") . parent::describeClass(parent::getVoc() . "Reference", "Reference") . parent::triplifyString($pmid, parent::getVoc() . "citation", $this->getString('./citation', $reference)) . parent::triplify($study_id, parent::getVoc() . str_replace("_", "-", $ref_type), $pmid));
}
}
}
} catch (Exception $e) {
echo "There was an error parsing references element: {$e}\n";
}
#######################################################################
#link
#######################################################################
try {
$links = $root->xpath('//link');
foreach ($links as $i => $link) {
$url = $this->getString('./url', $link);
$url = preg_replace("/>.*\$/", "", $url);
$lid = parent::getRes() . md5($url);
parent::addRDF(parent::describeIndividual($lid, $this->getString('./description', $link), parent::getVoc() . "Link") . parent::describeClass(parent::getVoc() . "Link", "Link") . parent::triplify($lid, parent::getVoc() . "url", $url) . parent::triplify($study_id, parent::getVoc() . "link", $lid));
}
} catch (Exception $e) {
echo "There was an error parsing link element: {$e}\n";
}
############################################################################
#responsible party
############################################################################
try {
$rp = @array_shift($root->xpath('//responsible_party'));
if ($rp) {
$rp_id = parent::getRes() . md5($rp->asXML());
$label = $this->getString('./name_title', $rp);
if (!$label) {
$label = $this->getString('./organization', $rp);
} else {
$label .= ", " . $this->getString('./organization', $rp);
}
if (!$label) {
$label = $this->getString('./party_type', $rp);
}
$org_id = parent::getRes() . md5($this->getString('./organization', $rp));
parent::addRDF(parent::describeIndividual($rp_id, $label, parent::getVoc() . "Responsible-Party") . parent::describeClass(parent::getVoc() . "Responsible-Party", "Responsible Party") . parent::triplify($study_id, parent::getVoc() . "responsible-party", $rp_id) . parent::triplify($rp_id, parent::getVoc() . "organization", $org_id) . parent::describeIndividual($org_id, $this->getString('./organization', $rp), parent::getVoc() . "Organization") . parent::describeClass(parent::getVoc() . "Organization", "Organization") . parent::triplifyString($rp_id, parent::getVoc() . "name-title", $this->getString('./name_title', $rp)) . parent::triplifyString($rp_id, parent::getVoc() . "party-type", $this->getString('./party_type', $rp)) . parent::triplifyString($rp_id, parent::getVoc() . "investigator-affiliation", $this->getString('./investigator_affiliation', $rp)) . parent::triplifyString($rp_id, parent::getVoc() . "investigator-full-name", $this->getString('./investigator_full_name', $rp)) . parent::triplifyString($rp_id, parent::getVoc() . "investigator-title", $this->getString('./investigator_title', $rp)));
}
} catch (Exception $e) {
echo "There was an error parsing the responsible_party element: {$e}\n";
}
##############################################################################
# keywords
##############################################################################
try {
$keywords = $root->xpath('//keyword');
foreach ($keywords as $keyword) {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "keyword", (string) $keyword));
}
} catch (Exception $e) {
echo "There was an error parsing the keywords element: {$e}";
}
# mesh terms
# note: mesh terms are assigned using an imperfect algorithm
try {
$mesh_terms = $root->xpath('//condition_browse/mesh_term');
foreach ($mesh_terms as $mesh_term) {
$term = (string) $mesh_term;
$mesh_id = parent::getRes() . md5($term);
parent::addRDF(parent::triplify($study_id, parent::getVoc() . "condition-mesh", $mesh_id));
parent::addRDF(parent::triplifyString($mesh_id, "rdfs:label", $term));
}
} catch (Exception $e) {
echo "There was an error in mesh_terms: {$e}\n";
}
################################################################################
# regulated by fda? is section 801? has expanded access?
################################################################################
try {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "is-fda-regulated", $this->getString('is_fda_regulated')) . parent::triplifyString($study_id, parent::getVoc() . "is-section-801", $this->getString('is_section_801')) . parent::triplifyString($study_id, parent::getVoc() . "has-expanded-access", $this->getString('has_expanded_access')));
} catch (Exception $e) {
echo "There was an error parsing the is_fda_regulated element: {$e}\n";
}
###############################################################################
# mesh terms for the intervention browse
###############################################################################
try {
$a = array("condition_browse", "intervention_browse");
foreach ($a as $browse_type) {
$terms = $root->xpath("//{$browse_type}/mesh_term");
foreach ($terms as $term) {
$term_label = (string) $term;
$term_id = parent::getRes() . md5($term);
parent::addRDF(parent::describeIndividual($term_id, $term_label, parent::getVoc() . "Term") . parent::describeClass(parent::getVoc() . "Term", "Term") . parent::triplify($study_id, parent::getVoc() . str_replace("_", "-", $browse_type), $term_id));
}
}
} catch (Exception $e) {
echo "There was an error parsing {$browse_type}/mesh_term element: {$e}\n";
}
################################################################################
# clinical results
################################################################################
try {
$cr = @array_shift($root->xpath('//clinical_results'));
if ($cr) {
$cr_id = parent::getRes() . md5($study_id . $cr->asXML());
parent::addRDF(parent::describeIndividual($cr_id, "clinical results for {$study_id}", parent::getVoc() . "Clinical-Result") . parent::describeClass(parent::getVoc() . "Clinical-Result", "Clinical Result") . parent::triplifyString($cr_id, parent::getVoc() . "description", $this->getString('./desc', $cr)) . parent::triplifyString($cr_id, parent::getVoc() . "restrictive-agreement", $this->getString('./restrictive_agreement', $cr)) . parent::triplifyString($cr_id, parent::getVoc() . "limitations-and-caveats", $this->getString('./limitations_and_caveats', $cr)) . parent::triplify($study_id, parent::getVoc() . "clinical-result", $cr_id));
}
} catch (Exception $e) {
echo "There was an error parsing clinical results: {$e}\n";
}
################################################################################
# Participant Flow
################################################################################
try {
$pc = 1;
$mc = 1;
$wc = 1;
$pf = @array_shift($root->xpath('//clinical_results/participant_flow'));
if ($pf) {
$pf_id = parent::getRes() . md5($pf->asXML());
parent::addRDF(parent::describeIndividual($pf_id, "participant flow for {$study_id}", parent::getVoc() . "Participant-Flow") . parent::describeClass(parent::getVoc() . "Participant-Flow", "Participant-Flow") . parent::triplify($study_id, parent::getVoc() . "participant-flow", $pf_id) . parent::triplifyString($pf_id, parent::getVoc() . "recruitment-details", $this->getString('./recruitment_details', $pf)) . parent::triplifyString($pf_id, parent::getVoc() . "pre-assignment-details", $this->getString('./pre_assignment_details', $pf)));
$groups = @array_shift($pf->xpath('./group_list'));
foreach ($groups as $group) {
parent::addRDF(parent::triplify($pf_id, parent::getVoc() . "group", $this->makeGroup($group)));
}
//period_list
$periods = @array_shift($pf->xpath('./period_list'));
foreach ($periods as $period) {
$period_id = parent::getRes() . $nct_id . "/period/" . $pc++;
$period_title = $this->getString('./title', $period);
parent::addRDF(parent::describeIndividual($period_id, $period_title . " for {$nct_id}", parent::getVoc() . "Period") . parent::describeClass(parent::getVoc() . "Period", "Period") . parent::triplify($pf_id, parent::getVoc() . "period", $period_id));
// milestones
$milestones = @array_shift($period->xpath('./milestone_list'));
if ($milestones) {
foreach ($milestones as $milestone) {
$milestone_id = parent::getRes() . $nct_id . "/milestone/" . $mc++;
$label = $this->getString('./title', $milestone);
parent::addRDF(parent::describeIndividual($milestone_id, $label, parent::getVoc() . "Milestone") . parent::describeClass(parent::getVoc() . "Milestone", "Milestone") . parent::triplify($period_id, parent::getVoc() . "milestone", $milestone_id));
// participants
$p = 1;
$ps_list = @array_shift($milestone->xpath('./participants_list'));
foreach ($ps_list as $ps) {
$ps_id = $milestone_id . "/p/" . $p++;
$group_id = parent::getRes() . $this->nct_id . "/group/" . $ps->attributes()->group_id;
$count = (string) $ps->attributes()->count;
parent::addRDF(parent::describeIndividual($ps_id, "participant counts in " . $ps->attributes()->group_id . " for milestone {$mc} of {$nct_id}", parent::getVoc() . "Participant-Count") . parent::describeClass(parent::getVoc() . "Participant-Count", "Participant Count") . parent::triplify($ps_id, parent::getVoc() . "group", $group_id) . parent::triplifyString($ps_id, parent::getVoc() . "count", $count) . parent::triplify($milestone_id, parent::getVoc() . "participant-counts", $ps_id));
}
}
}
// milestones
$withdraws = @array_shift($period->xpath('./drop_withdraw_reason_list'));
if ($withdraws) {
foreach ($withdraws as $withdraw) {
$wid = parent::getRes() . $this->nct_id . "/withdraw/" . $wc++;
$label = $this->getString('./title', $withdraw);
parent::addRDF(parent::describeIndividual($wid, $label, parent::getVoc() . "Withdraw-Reason") . parent::describeClass(parent::getVoc() . "Withdraw-Reason", "Withdraw Reason"));
// participants
$ps_list = @array_shift($withdraw->xpath('./participants_list'));
foreach ($ps_list as $ps) {
$group_id = parent::getRes() . $nct_id . "/group/" . $ps->attributes()->group_id;
$count = (string) $ps->attributes()->count;
parent::addRDF(parent::triplify($wid, parent::getVoc() . "group", $group_id) . parent::triplifyString($wid, parent::getVoc() . "count", $count));
}
}
}
}
}
} catch (Exception $e) {
echo "There was an error parsing participant flow element: {$e}\n";
}
################################################################################
# baseline
################################################################################
try {
$baseline = @array_shift($root->xpath('//baseline'));
if ($baseline) {
$b_id = $this->nct_id . "/baseline";
$b_uri = parent::getRes() . $b_id;
// group list
$groups = @array_shift($baseline->xpath('./group_list'));
foreach ($groups as $group) {
parent::addRDF(parent::describeIndividual($b_uri, "baseline for {$nct_id}", parent::getVoc() . "Baseline") . parent::describeClass(parent::getVoc() . "Baseline", "Baseline") . parent::triplify($b_uri, parent::getVoc() . "group", $this->makeGroup($group)) . parent::triplify($study_id, parent::getVoc() . "baseline", $b_uri));
}
// measure list
$measures = @array_shift($baseline->xpath('./measure_list'));
foreach ($measures as $measure) {
parent::addRDF(parent::triplify($b_uri, parent::getVoc() . "measure", $this->makeMeasure($measure)));
}
}
} catch (Exception $e) {
echo "Error in parsing baseline" . PHP_EOL;
}
################################################################################
# outcomes
################################################################################
try {
$outcomes = @array_shift($root->xpath('//outcome_list'));
if ($outcomes) {
foreach ($outcomes as $i => $outcome) {
$outcome_id = $this->nct_id . "/outcome/" . ($i + 1);
$outcome_uri = parent::getRes() . $outcome_id;
$outcome_label = $this->getString("./title", $outcome);
if (!$outcome_label) {
$outcome_label = "outcome for " . $this->nct_id;
}
parent::addRDF(parent::describeIndividual($outcome_uri, $outcome_label, parent::getVoc() . "Outcome", $this->getString("./description", $outcome)) . parent::describeClass(parent::getVoc() . "Outcome", "Outcome") . parent::triplify($study_id, parent::getVoc() . "outcome", $outcome_uri) . parent::triplifyString($outcome_uri, parent::getVoc() . "type", $this->getString("./type", $outcome)) . parent::triplifyString($outcome_uri, parent::getVoc() . "time-frame", $this->getString("./time_frame", $outcome)) . parent::triplifyString($outcome_uri, parent::getVoc() . "safety-issue", $this->getString("./safety_issue", $outcome)) . parent::triplifyString($outcome_uri, parent::getVoc() . "posting-date", $this->getString("./posting-date", $outcome)) . parent::triplifyString($outcome_uri, parent::getVoc() . "population", $this->getString("./population", $outcome)));
$groups = @array_shift($outcome->xpath('./group_list'));
if ($groups) {
foreach ($groups as $group) {
parent::addRDF(parent::triplify($outcome_uri, parent::getVoc() . "group", $this->makeGroup($group)));
}
}
// measure list
$measures = @array_shift($outcome->xpath('./measure_list'));
if ($measures) {
foreach ($measures as $measure) {
parent::addRDF(parent::triplify($outcome_uri, parent::getVoc() . "measure", $this->makeMeasure($measure)));
}
}
// analysis list
$analyses = @array_shift($outcome->xpath('./analysis_list'));
if ($analyses) {
foreach ($analyses as $analysis) {
parent::addRDF(parent::triplify($outcome_uri, parent::getVoc() . "analysis", $this->makeAnalysis($analysis)));
}
}
}
}
} catch (Exception $e) {
echo "Error in parsing outcomes" . PHP_EOL;
}
################################################################################
# events
################################################################################
try {
$c_ev = $c_c = 1;
$reported_events = @array_shift($root->xpath('//reported_events'));
if ($reported_events) {
$rp_id = parent::getRes() . md5($reported_events->asXML());
$groups = @array_shift($reported_events->xpath('./group_list'));
parent::addRDF(parent::describeIndividual($rp_id, "Reported events for {$nct_id}", parent::getVoc() . "Reported-Events") . parent::describeClass(parent::getVoc() . "Reported-Events", "Reported Events") . parent::triplify($study_id, parent::getVoc() . "reported-events", $rp_id));
foreach ($groups as $group) {
parent::addRDF(parent::triplify($rp_id, parent::getVoc() . "group", $this->makeGroup($group)));
}
// events
$event_list = array("serious_events" => "Serious Event", "other_events" => "Other Event");
foreach ($event_list as $ev => $ev_label) {
$et = @array_shift($reported_events->xpath('./' . $ev));
if (!$et) {
continue;
}
$ev_uri = parent::getVoc() . str_replace(" ", "-", $ev_label);
$categories = @array_shift($et->xpath('./category_list'));
foreach ($categories as $category) {
$major_title = $this->getString('./title', $category);
$major_title_uri = parent::getRes() . md5($major_title);
$events = @array_shift($category->xpath('./event_list'));
foreach ($events as $event) {
$e_uri = parent::getRes() . $this->nct_id . "/{$ev}/" . $c_ev++;
$subtitle = (string) $this->getString('./sub_title', $event) . " for " . $this->nct_id;
$subtitle_uri = parent::getRes() . md5($subtitle);
parent::addRDF(parent::describeIndividual($e_uri, $subtitle, $ev_uri) . parent::describeClass($ev_uri, $ev_label) . parent::triplify($e_uri, parent::getVoc() . "sub-title", $subtitle_uri) . parent::describeIndividual($subtitle_uri, $subtitle, parent::getVoc() . "Event") . parent::describeClass(parent::getVoc() . "Event", "Event") . parent::triplify($e_uri, parent::getVoc() . "major-title", $major_title_uri) . parent::describeClass($major_title_uri, $major_title) . parent::triplify($rp_id, parent::getVoc() . str_replace("_", "-", $ev), $e_uri));
$counts = $event->xpath('./counts');
foreach ($counts as $c) {
$group_id = $c->attributes()->group_id;
$group_uri = parent::getRes() . $nct_id . "/group/" . $group_id;
$c_uri = $e_uri . "/count/" . $c_c++;
parent::addRDF(parent::describeIndividual($c_uri, $subtitle . " for " . $group_id . " in " . $this->nct_id, parent::getVoc() . "Event-Count") . parent::describeClass(parent::getVoc() . "Event-Count", "Event Count") . parent::triplify($c_uri, parent::getVoc() . "group", $group_uri) . parent::triplify($e_uri, parent::getVoc() . "count", $c_uri) . parent::triplifyString($c_uri, parent::getVoc() . "default-vocabulary", $this->getString('./default_vocab', $et)) . parent::triplifyString($c_uri, parent::getVoc() . "frequency-threshold", $this->getString('./frequency_threshold', $et)) . parent::triplifyString($c_uri, parent::getVoc() . "default-assessment", $this->getString('./default_assessment', $et)) . parent::triplifyString($c_uri, parent::getVoc() . "number-events", $c->attributes()->events) . parent::triplifyString($c_uri, parent::getVoc() . "subjects-affected", $c->attributes()->subjects_affected) . parent::triplifyString($c_uri, parent::getVoc() . "subjects-at-risk", $c->attributes()->subjects_at_risk));
}
}
}
}
}
} catch (Exception $e) {
echo "Error in parsing reported events" . PHP_EOL;
}
parent::writeRDFBufferToWriteFile();
}
$this->setCheckPoint('record');
$this->setCheckPoint('dataset');
}
/**
* parse an individual pubchem substance file
**/
function parse_substance_file($indir, $file)
{
$xml = new CXML($indir, $file);
while ($xml->Parse("PC-Substance") == TRUE) {
parent::setCheckpoint('record');
$this->parse_substance_record($xml);
}
}
function Parse($ldir, $infile)
{
$i = 0;
$xml = new CXML($ldir, $infile);
while ($xml->Parse("partner") == TRUE) {
$this->ParsePartnerEntry($xml);
$this->WriteRDFBufferToWriteFile();
//if($i++ == 10) break;
}
unset($xml);
$xml = new CXML($ldir, $infile);
while ($xml->Parse("drug") == TRUE) {
$this->ParseDrugEntry($xml);
$this->WriteRDFBufferToWriteFile();
//if($i++ == 10) break;
}
unset($xml);
}
/**
* process a results xml file from the download directory
**/
function process_result($infile, $curr_block)
{
$indir = $this->GetParameterValue('indir');
$outfile = $this->GetParameterValue("outdir");
$this->CreateDirectory($outfile);
$outfile .= basename($infile, ".xml") . ".nt";
$gz = false;
if ($this->GetParameterValue('gzip')) {
$outfile .= basename($infile, ".xml") . ".nq";
}
if ($this->GetParameterValue('gzip')) {
$outfile .= '.gz';
$gz = true;
}
$this->SetWriteFile($outfile, $gz);
$xml = new CXML($indir, basename($infile));
while ($xml->Parse("clinical_study") == TRUE) {
$root = $xml->GetXMLRoot();
#########################################################################################
# study ids
#########################################################################################
$nct_id = @array_shift($root->xpath("//id_info/nct_id"));
$org_study_id = @array_shift($root->xpath("//id_info/org_study_id"));
$secondary_id = @array_shift($root->xpath("//id_info/secondary_id"));
$study_id = "clinicaltrials:" . $nct_id;
$this->AddRDF($this->QQuad($study_id, "rdf:type", "clinicaltrials_vocabulary:Clinical-Study"));
$this->AddRDF($this->QQuadl($study_id, "dc:identifier", $nct_id));
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:org-study-identifier", $org_study_id));
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:secondary-identifier", $secondary_id));
##########################################################################################
#brief trial
##########################################################################################
$brief_title = @array_shift($root->xpath("//brief_title"));
if ($brief_title != "") {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:brief-title", $this->SafeLiteral($brief_title)));
$this->AddRDF($this->QQuadl($study_id, "rdfs:label", $this->SafeLiteral($brief_title) . " [{$study_id}]"));
}
##########################################################################################
#official title
##########################################################################################
$official_title = @array_shift($root->xpath("//official_title"));
if ($official_title != "") {
$this->AddRDF($this->QQuadl($study_id, "dc:title", $this->SafeLiteral($official_title)));
if (!$brief_title) {
$this->AddRDF($this->QQuadl($study_id, "rdfs:label", $this->SafeLiteral($official_title) . " [{$study_id}]"));
}
}
#########################################################################################
#acronym
#########################################################################################
$acronym = @array_shift($root->xpath("//acronym"));
if ($acronym != "") {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:acronym", $this->SafeLiteral($acronym)));
}
########################################################################################
#lead_sponser
########################################################################################
try {
$lead_sponsor = @array_shift($root->xpath('//sponsors/lead_sponsor'));
$agency = @array_shift($lead_sponsor->xpath("//agency"));
$agency_class = @array_shift($lead_sponsor->xpath("//agency_class"));
$lead_sponsor_id = "clinicaltrials_resource:" . md5($lead_sponsor->asXML());
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:lead-sponsor", $lead_sponsor_id));
$this->AddRDF($this->QQuad($lead_sponsor_id, "rdf:type", "clinicaltrials_vocabulary:Lead-Sponsor"));
$this->AddRDF($this->QQuadl($lead_sponsor_id, "dc:title", $this->SafeLiteral($agency)));
$this->AddRDF($this->QQuadl($lead_sponsor_id, "rdfs:label", $this->SafeLiteral($agency) . " [{$lead_sponsor_id}]"));
$this->AddRDF($this->QQuadl($lead_sponsor_id, "clinicaltrials_vocabulary:agency-class", $this->SafeLiteral($agency_class)));
} catch (Exception $e) {
echo "There was an error in the lead sponsor element: {$e}\n";
}
######################################################################################
#oversight info
######################################################################################
try {
$over_site = @array_shift($root->xpath('//oversight_info'));
$authority = @array_shift($over_site->xpath('//authority'));
$os_id = "clinicaltrials_resource:" . md5($over_site->asXML());
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:oversight-authority", $os_id));
$this->AddRDF($this->QQuad($os_id, "rdf:type", "clinicaltrials_vocabulary:Oversight-Authority"));
$this->AddRDF($this->QQuadl($os_id, "rdfs:label", $authority . " [{$os_id}]"));
$this->AddRDF($this->QQuadl($os_id, "dc:title", $authority));
} catch (Exception $e) {
echo "There was an error in the oversight info element: {$e}\n";
}
####################################################################################
# has_dmc
####################################################################################
$has_dmc = @array_shift($over_site->xpath('//has_dmc'));
if ($has_dmc != "") {
$this->AddRDF($this->QQuadl($os_id, "clinicaltrials_vocabulary:dmc", $has_dmc));
// what's a dmc?
}
###################################################################################
#brief summary
###################################################################################
$brief_summary = @array_shift($root->xpath('//brief_summary/textblock'));
if ($brief_summary) {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:brief-summary", $this->SafeLiteral($brief_summary)));
}
####################################################################################
# detailed description
####################################################################################
$detailed_description = @array_shift($root->xpath('//detailed_description/textblock'));
if ($detailed_description) {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:detailed-description", $this->SafeLiteral($detailed_description)));
}
#################################################################################
# overall status
#################################################################################
$overall_status = @array_shift($root->xpath('//overall_status'));
if ($overall_status) {
$status_id = "clinicaltrials_resource:" . md5($overall_status);
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:overall-status", $status_id));
$this->AddRDF($this->QQuad($status_id, "rdf:type", "clinicaltrials_vocabulary:Status"));
$this->AddRDF($this->QQuadL($status_id, "rdfs:label", $overall_status));
}
##################################################################################
# start date
##################################################################################
$start_date = @array_shift($root->xpath('//start_date'));
if ($start_date) {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:start-date", $this->SafeLiteral($start_date)));
}
###################################################################################
# completion date
##################################################################################
$completion_date = @array_shift($root->xpath('//completion_date'));
if ($completion_date != "") {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:completion-date", $this->SafeLiteral($completion_date)));
}
####################################################################################
# primary completion date
###################################################################################
$primary_completion_date = @array_shift($root->xpath('//primary_completion_date'));
if ($primary_completion_date != "") {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:primary-completion-date", $this->SafeLiteral($primary_completion_date)));
}
####################################################################################
# study type
####################################################################################
$study_type = @array_shift($root->xpath('//study_type'));
if ($study_type != "") {
$study_type_id = "clinicaltrials_resource:" . md5($study_type);
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:study-type", $study_type_id));
$this->AddRDF($this->QQuad($study_type_id, "rdf:type", "clinicaltrials_vocabulary:Study-Type"));
$this->AddRDF($this->QQuadL($study_type_id, "rdfs:label", $this->SafeLiteral($study_type) . " [{$study_type_id}]"));
}
####################################################################################
# phase
####################################################################################
$phase = @array_shift($root->xpath('//phase'));
if ($phase && $phase != "N/A") {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:phase", $this->SafeLiteral($phase)));
}
###############################################################################
# study design
###############################################################################
$study_design = @array_shift($root->xpath('//study_design'));
if ($study_design) {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:study-design", $this->SafeLiteral($study_design)));
}
################################################################################
#primary outcome
###############################################################################
$primary_outcome = @array_shift($root->xpath('//primary_outcome'));
if ($primary_outcome) {
try {
$measure = @array_shift($root->xpath('//primary_outcome/measure'));
$time_frame = @array_shift($root->xpath('//primary_outcome/time_frame'));
$safety_issue = @array_shift($root->xpath('//primary_outcome/saftey_issue'));
$description = @array_shift($root->xpath('//primary_outcome/description'));
$po_id = "clinicaltrials_resource:" . md5($nct_id . $primary_outcome->asXML());
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:primary-outcome", $po_id));
$this->AddRDF($this->QQuad($po_id, "rdf:type", "clinicaltrials_vocabulary:Primary-Outcome"));
$this->AddRDF($this->QQuadl($po_id, "rdfs:label", $this->SafeLiteral($measure . " " . $time_frame) . " [{$po_id}]"));
$this->AddRDF($this->QQuadl($po_id, "clinicaltrials_vocabulary:measure", $this->SafeLiteral($measure)));
if ($description) {
$this->AddRDF($this->QQuadl($po_id, "dc:description", $this->SafeLiteral($description)));
}
if ($time_frame) {
$this->AddRDF($this->QQuadl($po_id, "clinicaltrials_vocabulary:time-frame", $this->SafeLiteral($time_frame)));
}
if ($safety_issue) {
$this->AddRDF($this->QQuadl($po_id, "clinicaltrials_vocabulary:safety-issue", $this->SafeLiteral($safety_issue)));
}
} catch (Exception $e) {
echo "There was an error parsing the primary outcome element: {$e} \n";
}
}
#################################################################################
#secondary outcome
#################################################################################
try {
$secondary_outcomes = $root->xpath('//secondary_outcome');
foreach ($secondary_outcomes as $secondary_outcome) {
$measure = @array_shift($secondary_outcome->xpath('//measure'));
$time_frame = @array_shift($secondary_outcome->xpath('//time_frame'));
$safety_issue = @array_shift($secondary_outcome->xpath('//safety_issue'));
$so_id = "clinicaltrials_resource:" . md5($nct_id . $secondary_outcome->asXML());
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:secondary-outcome", $so_id));
$this->AddRDF($this->QQuad($so_id, "rdf:type", "clinicaltrials_vocabulary:Secondary-Outcome"));
$this->AddRDF($this->QQuadl($so_id, "rdfs:label", $this->SafeLiteral($measure . " " . $time_frame) . "[{$so_id}]"));
$this->AddRDF($this->QQuadl($so_id, "clinicaltrials_vocabulary:measure", $this->SafeLiteral($measure)));
if ($time_frame) {
$this->AddRDF($this->QQuadl($so_id, "clinicaltrials_vocabulary:time-frame", $this->SafeLiteral($time_frame)));
}
if ($safety_issue) {
$this->AddRDF($this->QQuadl($so_id, "clinicaltrials_vocabulary:safety-issue", $this->SafeLiteral($safety_issue)));
}
}
} catch (Exception $e) {
"There was an exception parsing the secondary outcomes element: {$e}\n";
}
##############################################################################
#number of arms
##############################################################################
try {
$no_of_arms = @array_shift($root->xpath('//number_of_arms'));
if ($no_of_arms != "") {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:number-of-arms", $no_of_arms));
}
} catch (Exception $e) {
echo "There was an exception parsing the number of arms element: {$e}\n";
}
##############################################################################
#enrollment
##############################################################################
try {
$enrollment = @array_shift($root->xpath('//enrollment'));
if ($enrollment) {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:enrollment", $this->SafeLiteral($enrollment)));
}
} catch (Exception $e) {
echo "There was an exception parsing the enrollment element: {$e}\n";
}
###############################################################################
#condition
###############################################################################
try {
$conditions = $root->xpath('//condition');
foreach ($conditions as $condition) {
$mesh_label_id = "clinicaltrials_resource:" . md5($this->SafeLiteral($condition));
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:condition", $mesh_label_id));
$this->AddRDF($this->QQuadl($mesh_label_id, "rdfs:label", $this->SafeLiteral($condition)));
}
} catch (Exception $e) {
echo "There was an exception parsing condition element: {$e}\n";
}
################################################################################
# arm_group
################################################################################
try {
$arm_groups = $root->xpath('//arm_group');
foreach ($arm_groups as $arm_group) {
$arm_group_label = @array_shift($arm_group->xpath('./arm_group_label'));
$arm_group_type = ucfirst(str_replace(" ", "_", @array_shift($arm_group->xpath('./arm_group_type'))));
$description = @array_shift($arm_group->xpath('./description'));
$arm_group_id = "clinicaltrials_resource:" . md5($arm_group->asXML());
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:arm-group", $arm_group_id));
$this->AddRDF($this->QQuadl($arm_group_id, "rdfs:label", $this->SafeLiteral($arm_group_label) . "[{$arm_group_id}]"));
$this->AddRDF($this->QQuad($arm_group_id, "rdf:type", "clinicaltrials_vocabulary:" . $arm_group_type));
$this->AddRDF($this->QQuadl($arm_group_id, "rdfs:comment", $this->SafeLiteral($description)));
}
} catch (Exception $e) {
echo "There was an exception in arm groups: {$e}\n";
}
##############################################################################
#intervention
##############################################################################
try {
$interventions = $root->xpath('//intervention');
foreach ($interventions as $intervention) {
$intervention_name = @array_shift($intervention->xpath('./intervention_name'));
$intervention_type = ucfirst(str_replace(" ", "_", @array_shift($intervention->xpath('./intervention_type'))));
$intervention_id = "clinicaltrials_resource:" . md5($intervention->asXML());
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:intervention", $intervention_id));
$this->AddRDF($this->QQuad($intervention_id, "rdf:type", "clinicaltrials_vocabulary:" . $intervention_type));
$this->AddRDF($this->QQuadl($intervention_id, "rdfs:label", $this->SafeLiteral($intervention_name)));
$description = @array_shift($intervention->xpath('./description'));
if ($description != "") {
$this->AddRDF($this->QQuadl($intervention_id, "rdfs:comment", $this->SafeLiteral($description)));
}
}
} catch (Exception $e) {
echo "There was an error in interventions {$e}\n";
}
###############################################################################
#eligibility
################################################################################
try {
$eligibility = @array_shift($root->xpath('//eligibility'));
if ($eligibility != null) {
$eligibility_id = "clinicaltrials_resource:" . md5($eligibility->asXML());
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:eligibility", $eligibility_id));
$this->AddRDF($this->QQuad($eligibility_id, "rdf:type", "clinicaltrials_vocabulary:Eligibility"));
if ($criteria = @array_shift($eligibility->xpath('./criteria'))) {
$text = str_replace("\n\n", "", @array_shift($criteria->xpath('./textblock')));
$c = preg_split("/(Inclusion Criteria\\:|Exclusion Criteria\\:)/", $text);
//inclusion
if (isset($c[1])) {
$d = explode(" - ", $c[1]);
// the lists are separated by a hyphen
foreach ($d as $inclusion) {
$inc = trim($inclusion);
if ($inc != '') {
$inc_id = "clinicaltrials_resource:" . md5($inc);
$this->AddRDF($this->QQuad($eligibility_id, "clinicaltrials_vocabulary:inclusion-criteria", $inc_id));
$this->AddRDF($this->QQuad($inc_id, "rdf:type", "clinicaltrials_vocabulary:Inclusion-Criteria"));
$this->AddRDF($this->QQuadL($inc_id, "rdfs:label", trim($this->SafeLiteral($inc))));
}
}
}
//exclusion
if (isset($c[2])) {
$d = explode(" - ", $c[1]);
foreach ($d as $exclusion) {
$exc = trim($exclusion);
if ($exc != '') {
$exc_id = "clinicaltrials_resource:" . md5($exc);
$this->AddRDF($this->QQuad($eligibility_id, "clinicaltrials_vocabulary:exclusion-criteria", $exc_id));
$this->AddRDF($this->QQuad($exc_id, "rdf:type", "clinicaltrials_vocabulary:Exclusion-Criteria"));
$this->AddRDF($this->QQuadL($exc_id, "rdfs:label", trim($this->SafeLiteral($exc))));
}
}
}
}
if ($gender = @array_shift($eligibility->xpath('./gender'))) {
$this->AddRDF($this->QQuadl($eligibility_id, "clinicaltrials_vocabulary:gender", $gender));
}
if ($minimum_age = @array_shift($eligibility->xpath('./minimum_age'))) {
if ($minimum_age != 'N/A') {
$this->AddRDF($this->QQuadL($eligibility_id, "clinicaltrials_vocabulary:minimum-age", trim(str_replace("Years", "", $minimum_age))));
}
}
if ($maximum_age = @array_shift($eligibility->xpath('./maximum_age'))) {
if ($maximum_age != 'N/A') {
$this->AddRDF($this->QQuadL($eligibility_id, "clinicaltrials_vocabulary:maximum-age", trim(str_replace("Years", "", $maximum_age))));
}
}
if ($healthy_volunteers = @array_shift($eligibility->xpath('./healthy_volunteers'))) {
$this->AddRDF($this->QQuadl($eligibility_id, "clinicaltrials_vocabulary:healthy-volunteers", $healthy_volunteers));
}
if ($study_pop = @array_shift($eligibility->xpath('./study_pop'))) {
$this->AddRDF($this->QQuadl($eligibility_id, "clinicaltrials_vocabulary:study-population", $study_pop->xpath('./textblock')));
}
if ($sampling_method = @array_shift($eligibility->xpath('./sampling_method'))) {
$this->AddRDF($this->QQuadl($eligibility_id, "clinicaltrials_vocabulary:sampling-method", $sampling_method->xpath('./textblock')));
}
}
} catch (Exception $e) {
echo "There was an error in eligibility: {$e}\n";
}
######################################################################################
#overall official - the person in charge
#####################################################################################
try {
$overall_official = @array_shift($root->xpath('//overall_official'));
if ($overall_official) {
$overall_official = "clinicaltrials_resource:" . md5($overall_official->asXML());
$last_name = @array_shift($root->xpath('//overall_official/last_name'));
$role = @array_shift($root->xpath('//overall_official/role'));
$affiliation = @array_shift($root->xpath('//overall_official/affiliation'));
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:overall-official", $overall_official));
$this->AddRDF($this->QQuad($overall_official, "rdf:type", "clinicaltrials_vocabulary:Overall-Official"));
$this->AddRDF($this->QQuadl($overall_official, "clinicaltrials_vocabulary:lastname", $last_name));
$this->AddRDF($this->QQuadl($overall_official, "clinicaltrials_vocabulary:role", $role));
$this->AddRDF($this->QQuadl($overall_official, "clinicaltrials_vocabulary:affiliation", $this->SafeLiteral($affiliation)));
}
} catch (Exception $e) {
echo "There was an error parsing the overal_official: {$e}\n";
}
##############################################################
# location of facility doing the testing
##############################################################
try {
$location = @array_shift($root->xpath('//location'));
if ($location) {
$location_id = "clinicaltrials_resource:" . md5($location->asXML());
$title = @array_shift($location->xpath('//name'));
$facility = $location->xpath('./facility');
$address = @array_shift($location->xpath('//address'));
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:location", $location_id));
$this->AddRDF($this->QQuadl($location_id, "dc:title", $title));
$this->AddRDF($this->QQuadl($location_id, "rdfs:label", $title . " [{$location_id}]"));
$this->AddRDF($this->QQuad($location_id, "rdfs:type", "clinicaltrials_vocabulary:Location"));
if ($address && ($city = @array_shift($address->xpath('./city'))) != null) {
$this->AddRDF($this->QQuadl($location_id, "clinicaltrials_vocabulary:city", $city));
}
if ($address && ($state = @array_shift($address->xpath('./state'))) != null) {
$this->AddRDF($this->QQuadl($location_id, "clinicaltrials_vocabulary:state", $state));
}
if ($address && ($zip = @array_shift($address->xpath('./zip'))) != null) {
$this->AddRDF($this->QQuadl($location_id, "clinicaltrials_vocabulary:zipcode", $zip));
}
if ($address && ($country = @array_shift($address->xpath('./country'))) != null) {
$this->AddRDF($this->QQuadl($location_id, "clinicaltrials_vocabulary:country", $country));
}
}
} catch (Exception $e) {
echo "There was an error parsing location: {$e}" . "\n";
}
###################################################################
# group
###################################################################
try {
$groups = $root->xpath('//group');
foreach ($groups as $group) {
$group_id = "clinicaltrials_resource:" . md5($group->asXML());
$title = @array_shift($group->xpath('./title'));
$description = @array_shift($group->xpath('./description'));
$id = $group->attributes()->group_id;
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:group", $group_id));
$this->AddRDF($this->QQuad($group_id, "rdf:type", "clinicaltrials_vocabulary:Group"));
$this->AddRDF($this->QQuadl($group_id, "dc:title", $title));
$this->AddRDF($this->QQuadl($group_id, "rdfs:comment", $this->SafeLiteral($description)));
$this->AddRDF($this->QQuadl($group_id, "dc:identifier", $id));
}
} catch (Exception $e) {
echo "There was an exception parsing groups xml element: {$e}\n";
}
######################################################################
#results reference
######################################################################
try {
$references = $root->xpath('//reference');
foreach ($references as $reference) {
$p = @array_shift($reference->xpath('./PMID'));
if ($p) {
$pmid = "pubmed:{$p}";
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:reference", $pmid));
$this->AddRDF($this->QQuadl($pmid, "rdfs:comment", $this->SafeLiteral(@array_shift($reference->xpath('./citation')))));
}
}
} catch (Exception $e) {
echo "There was an error parsing references element: {$e}\n";
}
#######################################################################
#results reference
#######################################################################
try {
$results_references = $root->xpath('//results_reference');
foreach ($results_references as $result_reference) {
$p = @array_shift($result_reference->xpath('./PMID'));
if ($p) {
$pmid = "pubmed:" . $p;
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:results-reference", $pmid));
$this->AddRDF($this->QQuadl($pmid, "rdfs:comment", $this->SafeLiteral(@array_shift($result_reference->xpath('./citation')))));
}
}
} catch (Exception $e) {
echo "There was an error parsing results_references element: {$e}\n";
}
##########################################################################
#verification date
#########################################################################
try {
$verification_date = @array_shift($root->xpath('//verification_date'));
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:verification-date", $verification_date));
$lastchanged_date = @array_shift($root->xpath('//lastchanged_date'));
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:last-changed-date", $lastchanged_date));
$firstreceived_date = @array_shift($root->xpath('//firstreceived_date'));
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:first-received-date", $firstreceived_date));
} catch (Exception $e) {
echo "There was an error parsing the verification_date element: {$e}\n";
}
############################################################################
#responsible party
############################################################################
try {
$responsible_party = @array_shift($root->xpath('//responsible_party'));
if ($responsible_party) {
$name_title = $root->xpath('//responsible_party/name_title');
$organization = $root->xpath('//responsible_party/organization');
$rp_id = "clinicaltrials_resource:" . md5($responsible_party->asXML());
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:responsible-party", $rp_id));
$this->AddRDF($this->QQuad($rp_id, "rdf:type", "clinicaltrials_vocabulary:Responsible-Party"));
$this->AddRDF($this->QQuadl($rp_id, "rdfs:label", "{$name_title}, {$organization} [{$rp_id}]"));
$this->AddRDF($this->QQuadl($rp_id, "clinicaltrials_vocabulary:name-title", $name_title));
$this->AddRDF($this->QQuadl($rp_id, "clinicaltrials_vocabulary:organization", $organization));
}
} catch (Exception $e) {
echo "There was an error parsing the responsible_party element: {$e}\n";
}
##############################################################################
# key words
##############################################################################
try {
$keywords = $root->xpath('//keyword');
foreach ($keywords as $keyword) {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:keyword", $keyword));
}
} catch (Exception $e) {
echo "There was an error parsing the keywords element: {$e}";
}
# mesh terms
# note: mesh terms are assigned using an imperfect algorithm
try {
$mesh_terms = $root->xpath('//condition_browse/mesh_term');
foreach ($mesh_terms as $mesh_term) {
$mesh_id = "clinicaltrials_resource:" . md5($mesh_term);
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:condition-mesh", $mesh_id));
$this->AddRDF($this->QQuadl($mesh_id, "rdfs:label", $mesh_term));
}
} catch (Exception $e) {
echo "There was an error in mesh_terms: {$e}\n";
}
###############################################################################
# mesh terms for hte invervention browse
###############################################################################
try {
$mesh_terms = $root->xpath('//intervention_browse/mesh_term');
foreach ($mesh_terms as $mesh_label) {
$mesh_label_id = "clinicaltrials_resource:" . md5($mesh_label);
$this->AddRDF($this->QQuad($study_id, "clinicaltrials_vocabulary:intervention_mesh", $mesh_label_id));
$this->AddRDF($this->QQuadl($mesh_label_id, "rdfs:label", $this->SafeLiteral($mesh_label)));
}
} catch (Exception $e) {
echo "There was an error parsing intervention_browse/mesh_term element: {$e}\n";
}
################################################################################
# regulated by fda?
# boolean value yes or no
################################################################################
try {
$regulated = @array_shift($root->xpath('is_fda_regulated'));
if ($regulated != "") {
$this->AddRDF($this->QQuadl($study_id, "clinicaltrials_vocabulary:is-fda-regulated", $regulated));
}
} catch (Excepetion $e) {
echo "There was an error parsing the is_fda_regulated element: {$e}\n";
}
$this->WriteRDFBufferToWriteFile();
}
$this->getWriteFile()->close();
}
/**
* parse an individual pubchem substance file
**/
function parse_substance_file($indir, $file)
{
$xml = new CXML($indir, $file);
while ($xml->Parse("PC-Substance") == TRUE) {
$this->parse_substance_record($xml);
$this->WriteRDFBufferToWriteFile();
}
}
function parse_drugbank($ldir, $infile)
{
$xml = new CXML($ldir . $infile);
while ($xml->parse("drug") == TRUE) {
if (isset($this->id_list) and count($this->id_list) == 0) {
break;
}
$this->parseDrugEntry($xml);
}
unset($xml);
}
function genes($file)
{
$xml = new CXML($file);
while ($xml->parse("DisorderList") == TRUE) {
$x = $xml->GetXMLRoot();
foreach ($x->Disorder as $d) {
$orphanet_id = parent::getNamespace() . (string) $d->OrphaNumber;
$disorder_name = (string) $d->Name;
foreach ($d->DisorderGeneAssociationList->DisorderGeneAssociation as $dga) {
// gene
$gene = $dga->Gene;
$gene_id = parent::getNamespace() . (string) $gene->OrphaNumber;
$gene_internal_id = (string) $gene->attributes()->id;
$gene_label = (string) $gene->Name;
$gene_symbol = (string) $gene->Symbol;
parent::addRDF(parent::describeIndividual($gene_id, $gene_label, parent::getVoc() . "Gene") . parent::describeClass(parent::getVoc() . "Gene", "orphanet gene") . parent::triplifyString($gene_id, parent::getVoc() . "symbol", $gene_symbol));
foreach ($gene->SynonymList as $s) {
$synonym = (string) $s->Synonym;
parent::addRDF(parent::triplifyString($gene_id, parent::getVoc() . "synonym", $synonym));
}
foreach ($gene->ExternalReferenceList as $erl) {
$er = $erl->ExternalReference;
$db = (string) $er->Source;
$db = parent::getRegistry()->getPreferredPrefix($db);
$id = (string) $er->Reference;
$xref = "{$db}:{$id}";
parent::addRDF(parent::triplify($gene_id, parent::getVoc() . "x-{$db}", $xref));
}
$dga_id = parent::getRes() . (string) $d->OrphaNumber . "_" . md5($dga->asXML());
$ga = $dga->DisorderGeneAssociationType;
$ga_id = parent::getNamespace() . (string) $ga->attributes()->id;
$ga_label = (string) $ga->Name;
$s = $dga->DisorderGeneAssociationStatus;
$s_id = parent::getNamespace() . (string) $s->attributes()->id;
$s_label = (string) $s->Name;
parent::addRDF(parent::describeIndividual($dga_id, "{$ga_label} {$gene_label} in {$disorder_name} ({$s_label})", $ga_id) . parent::describeClass($ga_id, $ga_label, parent::getVoc() . "Disorder-Gene-Association") . parent::triplify($dga_id, parent::getVoc() . "status", $s_id) . parent::describeClass($s_id, $s_label, parent::getVoc() . "Disorder-Gene-Association-Status") . parent::triplify($dga_id, parent::getVoc() . "disorder", $orphanet_id) . parent::describeIndividual($orphanet_id, $disorder_name, parent::getVoc() . "Disorder") . parent::triplify($dga_id, parent::getVoc() . "gene", $gene_id));
}
parent::writeRDFBufferToWriteFile();
}
}
unset($xml);
}
function parse($file)
{
$xml = new CXML($file);
$xml->parse();
$entry = $xml->getXMLRoot();
if (!isset($entry) or !$entry) {
return false;
}
foreach ($entry->children() as $o) {
$rsid = "rs" . $o->attributes()->rsId;
$id = parent::getNamespace() . $rsid;
$type = parent::getVoc() . ucfirst(str_replace(" ", "-", (string) $o->attributes()->snpClass));
$snpclass = parent::getVoc() . (string) $o->attributes()->snpClass;
$moltype = parent::getVoc() . (string) $o->attributes()->molType;
// attributes
parent::addRDF(parent::describeIndividual($id, $rsid, $type) . parent::describeClass($type, ucfirst("" . $o->attributes()->snpClass)) . parent::triplify($id, parent::getVoc() . "mol-type", $moltype) . parent::describeClass($moltype, (string) $o->attributes()->molType, parent::getVoc() . "Moltype") . parent::describeClass(parent::getVoc() . "Moltype", "Moltype") . parent::triplify($id, parent::getVoc() . "taxid", "taxonomy:" . (string) $o->attributes()->taxId));
$genotype = (string) $o->attributes()->genoType;
if ($genotype) {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "genotype", parent::getVoc() . $genotype, "xsd:bool"));
}
// frequency
// create/update
/* if(!isset($o->Update)) $a = $o->Create;
else $a = $o->Update;
parent::addRDF(parent::triplifyString($id,parent::getVoc()."build",(string) $a->attributes()->build));
*/
//validation
$a = $o->Validation;
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "validation-by-cluster", (string) $a->attributes()->byCluster) . parent::triplifyString($id, parent::getVoc() . "validation-by-frequency", (string) $a->attributes()->byFrequency) . parent::triplifyString($id, parent::getVoc() . "validation-by-2hit2allele", (string) $a->attributes()->by2Hit2Allele) . parent::triplifyString($id, parent::getVoc() . "validation-by-1000G", (string) $a->attributes()->by1000G));
//hgvs names
foreach ($o->hgvs as $name) {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "hgvs-name", (string) $name));
}
// assembly
$assembly = $o->Assembly;
if ($assembly and $assembly->attributes()->reference == "true") {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "dbsnp-build", (string) $assembly->attributes()->dbSnpBuild) . parent::triplifyString($id, parent::getVoc() . "genome-build", (string) $assembly->attributes()->genomeBuild));
$component = $assembly->Component;
if ($component) {
parent::addRDF(parent::triplify($id, parent::getVoc() . "contig-accession", "genbank:" . (string) $component->attributes()->accession) . parent::triplify($id, parent::getVoc() . "contig-gi", "gi:" . (string) $component->attributes()->gi) . parent::triplifyString($id, parent::getVoc() . "chromosome", (string) $component->attributes()->chromosome));
$maploc = $component->MapLoc;
if ($maploc) {
foreach ($maploc->children() as $fxnset) {
$fxnset_id = parent::getRes() . md5($fxnset->asXML());
parent::addRDF(parent::triplify($id, parent::getVoc() . "maps-to", $fxnset_id) . parent::triplify($fxnset_id, "rdf:type", parent::getVoc() . "Fxnset") . parent::describeClass(parent::getVoc() . "Fxnset", "Fxnset"));
if (isset($fxnset->attributes()->geneId)) {
parent::addRDF(parent::triplify($fxnset_id, parent::getVoc() . "gene", "ncbigene:" . (string) $fxnset->attributes()->geneId));
}
if (isset($fxnset->attributes()->symbol)) {
parent::addRDF(parent::triplifyString($fxnset_id, parent::getVoc() . "gene-symbol", (string) $fxnset->attributes()->symbol));
}
if (isset($fxnset->attributes()->mrnaAcc)) {
parent::addRDF(parent::triplify($fxnset_id, parent::getVoc() . "mrna", "refseq:" . (string) $fxnset->attributes()->mrnaAcc));
}
if (isset($fxnset->attributes()->protAcc)) {
parent::addRDF(parent::triplify($fxnset_id, parent::getVoc() . "protein", "refseq:" . (string) $fxnset->attributes()->protAcc));
}
if (isset($fxnset->attributes()->fxnClass)) {
parent::addRDF(parent::triplifyString($fxnset_id, parent::getVoc() . "fxn-class", (string) $fxnset->attributes()->fxnClass));
}
if (isset($fxnset->attributes()->allele)) {
parent::addRDF(parent::triplifyString($fxnset_id, parent::getVoc() . "allele", (string) $fxnset->attributes()->allele));
}
if (isset($fxnset->attributes()->residue)) {
parent::addRDF(parent::triplifyString($fxnset_id, parent::getVoc() . "residue", (string) $fxnset->attributes()->residue));
}
if (isset($fxnset->attributes()->readingFrame)) {
parent::addRDF(parent::triplifyString($fxnset_id, parent::getVoc() . "reading-frame", (string) $fxnset->attributes()->readingFrame));
}
if (isset($fxnset->attributes()->aaPosition)) {
parent::addRDF(parent::triplifyString($fxnset_id, parent::getVoc() . "position", (string) $fxnset->attributes()->aaPosition));
}
}
}
}
}
}
unset($xml);
}
