/**
* 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');
$this->setCheckPoint('dataset');
}
function Parse($xml)
{
// state the dataset info
foreach ($xml->release->dbinfo as $o) {
$db = $o->attributes()->dbname . " v" . $o->attributes()->version . " (" . $o->attributes()->entry_count . " entries) [" . $o->attributes()->file_date . "]";
parent::addRDF(parent::triplifyString(parent::getDatasetURI(), parent::getVoc() . "contains", $db));
if ((string) $o->attributes()->dbname === "INTERPRO") {
parent::setDatasetVersion($o->attributes()->version);
}
}
// get a potential id list
if (parent::getParameterValue("id_list") != '') {
$id_list = explode(",", parent::getParameterValue("id_list"));
}
// now interate over the entries
foreach ($xml->interpro as $o) {
parent::writeRDFBufferToWriteFile();
$interpro_id = $o->attributes()->id;
if (isset($id_list) && !in_array($interpro_id, $id_list)) {
continue;
}
echo "Processing {$interpro_id}" . PHP_EOL;
$name = $o->name;
$short_name = $o->attributes()->short_name;
$type = $o->attributes()->type;
$s = parent::getNamespace() . $interpro_id;
//echo "Adding... $s rdfs:label $name ($short_name) $type [$s]".PHP_EOL;
parent::addRDF(parent::describeIndividual($s, "{$name} ({$short_name}) {$type}", parent::getVoc() . $type));
// get the pubs
unset($pubs);
foreach ($o->pub_list->publication as $p) {
$pid = (string) $p->attributes()->id;
if (isset($p->db_xref)) {
if ($p->db_xref->attributes()->db == "PUBMED") {
$pmid = (string) $p->db_xref->attributes()->dbkey;
$pubs['pid'][] = '<cite idref="' . $pid . '"/>';
$pubs['pmid'][] = '<a href="http://www.ncbi.nlm.nih.gov/pubmed/' . $pmid . '">pubmed:' . $pmid . '</a>';
parent::addRDF(parent::triplify($s, parent::getVoc() . "x-pubmed", "pubmed:{$pmid}"));
}
}
}
$abstract = (string) $o->abstract->p->asXML();
if (isset($pubs)) {
$abstract = str_replace($pubs['pid'], $pubs['pmid'], $abstract);
}
parent::addRDF(parent::triplifyString($s, "dc:description", $abstract));
if (isset($o->example_list)) {
foreach ($o->example_list->example as $example) {
$db = (string) $example->db_xref->attributes()->db;
$id = (string) $example->db_xref->attributes()->dbkey;
parent::addRDF(parent::triplify($s, parent::getVoc() . "example-entry", "{$db}:{$id}"));
}
}
if (isset($o->parent_list->rel_ref)) {
foreach ($o->parent_list->rel_ref as $parent) {
$id = (string) $parent->attributes()->ipr_ref;
parent::addRDF(parent::triplify($s, parent::getVoc() . "parent", "interpro:{$id}"));
}
}
if (isset($o->child->rel_ref)) {
foreach ($o->child->rel_ref as $child) {
$id = (string) $child->attributes()->ipr_ref;
parent::addRDF(parent::triplify($s, parent::getVoc() . "child", "interpro:{$id}"));
}
}
if (isset($o->contains->rel_ref)) {
foreach ($o->contains->rel_ref as $contains) {
$id = (string) $contains->attributes()->ipr_ref;
parent::addRDF(parent::triplify($s, parent::getVoc() . "contains", "interpro:{$id}"));
}
}
if (isset($o->found_in->rel_ref)) {
foreach ($o->found_in->rel_ref as $f) {
$id = (string) $f->attributes()->ipr_ref;
parent::addRDF(parent::triplify($s, parent::getVoc() . "found-in", "interpro:{$id}"));
}
}
if (isset($o->sec_list->sec_ac)) {
foreach ($o->sec_ac as $s) {
$id = (string) $s->attributes()->acc;
parent::addRDF(parent::triplify($s, parent::getVoc() . "secondary-accession", "interpro:{$id}"));
}
}
// xrefs
if (isset($o->member_list->dbxref)) {
foreach ($o->member_list->db_xref as $dbxref) {
$db = (string) $dbxref->attributes()->db;
$id = (string) $dbxref->attributes()->dbkey;
parent::addRDF(parent::triplify($s, parent::getVoc() . "x-" . strtolower($db), "{$db}:{$id}"));
}
}
if (isset($o->external_doc_list)) {
foreach ($o->external_doc_list->db_xref as $dbxref) {
$db = (string) $dbxref->attributes()->db;
$id = (string) $dbxref->attributes()->dbkey;
parent::addRDF(parent::triplify($s, parent::getVoc() . "x-" . strtolower($db), "{$db}:{$id}"));
}
}
if (isset($o->structure_db_links->db_xref)) {
foreach ($o->structure_db_links->db_xref as $dbxref) {
$db = (string) $dbxref->attributes()->db;
$id = (string) $dbxref->attributes()->dbkey;
parent::addRDF(parent::triplify($s, parent::getVoc() . "x-" . strtolower($db), "{$db}:{$id}"));
}
}
// taxon distribution
foreach ($o->taxonomy_distribution->taxon_data as $t) {
$organism = (string) $t->attributes()->name;
$number = (string) $t->attributes()->proteins_count;
parent::addRDF(parent::triplifyString($s, parent::getVoc() . "taxon-distribution", "{$organism} ({$number})"));
}
}
}
private function gene2accession()
{
$this->getReadFile()->read(200000);
$header = array(0 => array('rel' => "x-taxonomy", 'ns' => "taxonomy"), 1 => array('rel' => "ncbigene", 'ns' => "ncbigene"), 2 => array('rel' => "status"), 3 => array('rel' => "rna-nucleotide-accession.version", 'ns' => "genbank"), 4 => array('rel' => "rna-nucleotide-gi", 'ns' => "gi"), 5 => array('rel' => "protein-accession.version", 'ns' => "genbank"), 6 => array('rel' => "protein-gi", 'ns' => "gi"), 7 => array('rel' => "genomic-nucleotide-accession.version", 'ns' => "genbank"), 8 => array('rel' => "genomic-nucleotide-gi", 'ns' => "gi"), 9 => array('rel' => "genomic-start-position"), 10 => array('rel' => "genomic-end-position"), 11 => array('rel' => "orientation"), 12 => array('rel' => "assembly"), 13 => array('rel' => "mature-peptide-accession.version", 'ns' => "genbank"), 14 => array('rel' => "mature-peptide-gi", 'ns' => "gi"), 15 => array('rel' => "symbol"));
//(tab is used as a separator, pound sign - start of a comment) */
$z = 1;
while ($l = $this->getReadFile()->read(200000)) {
if ($l[0] == "#") {
continue;
}
if ($z++ % 10000 == 0) {
echo $z . PHP_EOL;
parent::clear();
}
$a = explode("\t", rtrim($l));
if (count($a) != 16) {
trigger_error("gene2accession: expecting 16 columns, found " . count($a) . " instead", E_USER_ERROR);
}
$taxid = $a[0];
if (isset($this->taxids) and !isset($this->taxids[$taxid])) {
continue;
}
$id = parent::getNamespace() . $a[1];
$refseq = false;
if ($a[2] != '-') {
$refseq = true;
}
if ($a[9] != '-' and $a[10] != '-') {
$region = parent::getRes() . $a[7] . "/" . $a[9] . "-" . $a[10];
$start_pos = parent::getRes() . $a[7] . "/" . $a[9];
$stop_pos = parent::getRes() . $a[7] . "/" . $a[10];
if ($a[11] == "+") {
$orientation = "faldo:ForwardStrandPosition";
} else {
if ($a[11] == "-") {
$orientation = "faldo:ReverseStrandPosition";
} else {
$orientation = "faldo:StrandedPosition";
}
}
parent::addRDF(parent::describeIndividual($region, "location of ncbigene:" . $a[1] . " on " . $a[7], "faldo:Region") . parent::describeIndividual($start_pos, "start of ncbigene:" . $a[1] . " on " . $a[7], "faldo:ExactPosition") . parent::describeIndividual($stop_pos, "stop position of ncbigene:" . $a[1] . " on " . $a[7], "faldo:ExactPosition") . parent::triplify($id, "faldo:location", $region) . parent::triplify($region, "faldo:begin", $start_pos) . parent::triplify($start_pos, "rdf:type", $orientation) . parent::triplifyString($start_pos, "faldo:position", $a[9], "xsd:integer") . parent::triplify($start_pos, "faldo:reference", "refseq:" . $a[7]) . parent::triplify($region, "faldo:end", $stop_pos) . parent::triplify($stop_pos, "rdf:type", $orientation) . parent::triplifyString($stop_pos, "faldo:position", $a[10], "xsd:integer") . parent::triplify($stop_pos, "faldo:reference", "refseq:" . $a[7]));
}
foreach ($header as $i => $v) {
if ($a[$i] == "-") {
continue;
}
if ($i == 1 or $i == 9 or $i == 10 or $i == 11) {
continue;
}
/// ncbigene
if (isset($v['ns'])) {
$ns = $v['ns'];
if ($ns == 'genbank' and $refseq == true) {
$ns = 'refseq';
}
parent::addRDF(parent::triplify($id, parent::getVoc() . $v['rel'], "{$ns}:" . $a[$i]));
} else {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . $v['rel'], $a[$i]));
}
}
parent::writeRDFBufferToWriteFile();
}
//while
}
function gene_interactions()
{
while ($l = parent::getReadFile()->Read()) {
if ($l[0] == '#') {
continue;
}
$data = explode("\t", $l);
if (count($data) != 11) {
trigger_error("Found " . count($data) . " columns, expecting 11");
continue;
}
$interaction = $data[0];
$interaction_type = str_replace("_", "-", $data[1]);
$interaction_type_label = str_replace("_", " ", $data[1]);
$int_additional_info = $data[2];
$gene1 = $data[5];
$gene2 = $data[8];
$interaction_id = parent::getNamespace() . $interaction;
if ($interaction_type == "Genetic") {
$int_pred = parent::getVoc() . "genetically-interacts-with";
} elseif ($interaction_type == "Physical") {
$int_pred = parent::getVoc() . "physically-interacts-with";
} elseif ($interaction_type == "Predicted") {
$int_pred = parent::getVoc() . "predicted-to-interact-with";
} elseif ($interaction_type == "Regulatory") {
$int_pred = parent::getVoc() . "regulates";
}
//elseif
if ($int_additional_info == "No_interaction") {
$interaction_label = "No " . strtolower($interaction_type) . " interaction between " . $gene1 . " and " . $gene2;
parent::addRDF(parent::describeIndividual($interaction_id, $interaction_label, parent::getVoc() . $interaction_type . "-Non-Interaction") . parent::describeClass(parent::getVoc() . $interaction_type . "-Non-Interaction", $interaction_type_label . " non-interaction") . parent::triplify($interaction_id, parent::getVoc() . "involves", parent::getNamespace() . $gene1) . parent::triplify($interaction_id, parent::getVoc() . "involves", parent::getNamespace() . $gene2));
$npa_id = parent::getRes() . md5($interaction_id . "negative property assertion");
$npa_label = "Negative property assertion stating that " . $gene1 . " and " . $gene2 . " do not have a " . $interaction_type_label . " interaction";
parent::addRDF(parent::describeIndividual($npa_id, $npa_label, "owl:NegativeObjectPropertyAssertion") . parent::triplify($npa_id, "owl:sourceIndividual", parent::getNamespace() . $gene1) . parent::triplify($npa_id, "owl:targetIndividual", parent::getNamespace() . $gene2) . parent::triplify($npa_id, "owl:assertionProperty", $int_pred));
} elseif ($int_additional_info == "N/A" || $int_additional_info == "Genetic_interaction") {
$interaction_label = $interaction_type . " interaction between " . $gene1 . " and " . $gene2;
parent::addRDF(parent::describeIndividual($interaction_id, $interaction_label, parent::getVoc() . $interaction_type . "-Interaction") . parent::describeClass(parent::getVoc() . $interaction_type . "-Interaction", $interaction_type_label . " Interaction") . parent::triplify($interaction_id, parent::getVoc() . "involves", parent::getNamespace() . $gene1) . parent::triplify($interaction_id, parent::getVoc() . "involves", parent::getNamespace() . $gene2) . parent::triplify(parent::getNamespace() . $gene1, $int_pred, parent::getNamespace() . $gene2));
} else {
$interaction_label = ($int_additional_info != "" ? $int_additional_info . " " : "") . strtolower($interaction_type) . " interaction between " . $gene1 . " and " . $gene2;
$type = parent::getVoc() . ($int_additional_info != "" ? $int_additional_info . "-" : "") . $interaction_type . "-Interaction";
$type_label = ($int_additional_info != "" ? $int_additional_info . " " : "") . $interaction_type_label . " Interaction";
parent::addRDF(parent::describeIndividual($interaction_id, $interaction_label, $type) . parent::describeClass($type, $type_label, parent::getVoc() . $interaction_type . "-Interaction") . parent::describeClass(parent::getVoc() . $interaction_type . "-Interaction", $interaction_type . " Interation") . parent::triplify($interaction_id, parent::getVoc() . "involves", parent::getNamespace() . $gene1) . parent::triplify($interaction_id, parent::getVoc() . "involves", parent::getNamespace() . $gene2) . parent::triplify(parent::getNamespace() . $gene1, $int_pred, parent::getNamespace() . $gene2));
}
//else
parent::WriteRDFBufferToWriteFile();
}
//while
}
function pathways()
{
// needs to be finished
return;
while ($l = $this->getReadFile()->read(50000)) {
$a = explode("\t", trim($l));
// From To Reaction Type Controller Control Type Cell Type PubMed Id Genes Drugs Diseases
// hmg coa reductase inhibitors Active & Inactive metabolites Biochemical Reaction CYP2C19,CYP2C8,CYP2C9,CYP2D6,CYP3A4,CYP3A5,UGT1A1,UGT1A3,UGT2B7 Catalysis hepatocyte CYP3A4,CYP3A5,UGT1A3,CYP2C19,CYP2C9,CYP2C8,CYP2D6,UGT1A1,UGT2B7 hmg coa reductase inhibitors
$c1 = array_search($a[0], $this->drugs);
if ($c1 === FALSE) {
$c1 = array_search($a[0], $this->genes);
if ($c1 === FALSE) {
$c1 = parent::getRes() . url_encode($c1);
} else {
$c1 = parent::getNamespace() . $c1;
}
}
$c2 = array_search($a[1], $this->drugs);
if ($c2 === FALSE) {
$c2 = array_search($a[1], $this->genes);
if ($c2 === FALSE) {
// not found
$c2 = parent::getRes() . url_encode($c2);
} else {
// actual id
$c2 = parent::getNamespace() . $c2;
}
}
$id = md5($l);
$uri = parent::getRes() . $id;
parent::writeRDFBufferToWriteFile();
}
}
function product($fpin)
{
$z = 0;
$list = '';
fgets($fpin);
// header
while ($l = fgets($fpin, 100000)) {
$a = explode("\t", $l);
if (count($a) != 18) {
trigger_error("Expected 18 coloumns, instead found" . count($a));
continue;
}
$product_id = parent::getNamespace() . $a[0];
$product_label = $a[3];
$product_type_label = ucfirst(strtolower($a[2]));
$product_type = parent::getVoc() . str_replace(" ", "-", $product_label);
parent::addRDF(parent::describeIndividual($product_id, $a[3], parent::getVoc() . "Product") . parent::describeClass(parent::getVoc() . "Product", "NDC Product") . parent::triplify($product_id, parent::getVoc() . "product-type", $product_type) . parent::describeIndividual($product_type, $product_type_label, parent::getVoc() . "Product-Type") . parent::describeClass(parent::getVoc() . "Product-Type", "Product Type") . parent::triplifyString($product_id, parent::getVoc() . "product-id", $a[1]) . parent::triplifyString($product_id, parent::getVoc() . "proprietary-name", $a[3]) . parent::triplifyString($product_id, parent::getVoc() . "trade-name-suffix", $a[4]));
if ($a[5]) {
$b = explode(";", $a[5]);
foreach ($b as $c) {
parent::addRDF(parent::triplifyString($product_id, parent::getVoc() . "non-proprietary-name", trim($c)));
}
}
if ($a[6]) {
$b = explode(",", $a[6]);
foreach ($b as $c) {
$dosageform = strtolower($c);
$dosageform_id = parent::getVoc() . str_replace(" ", "-", ucfirst(strtolower($c)));
parent::addRDF(parent::describeIndividual($dosageform_id, $dosageform, parent::getVoc() . "Dosage-Form") . parent::describeClass(parent::getVoc() . "Dosage-Form", "NDC Dosage Form") . parent::triplify($product_id, parent::getVoc() . "dosage-form", $dosageform_id));
}
}
if ($a[7]) {
// MV
$b = explode("; ", $a[7]);
foreach ($b as $c) {
$route = strtolower(trim($c));
$route_id = parent::getVoc() . str_replace(" ", "-", ucfirst(strtolower($c)));
parent::addRDF(parent::describeIndividual($route_id, $route, parent::getVoc() . "Route") . parent::describeClass(parent::getVoc() . "Route", "NDC Drug Route") . parent::triplify($product_id, parent::getVoc() . "route", $route_id));
}
}
if ($a[8]) {
$date = substr(0, 4, $a[8]) . "-" . substr(4, 2, $a[8]) . "-" . substr(6, 2, $a[8]);
parent::addRDF(parent::triplifyString($product_id, parent::getVoc() . "start-marketing-date", $date));
}
if ($a[9]) {
$date = substr(0, 4, $a[9]) . "-" . substr(4, 2, $a[9]) . "-" . substr(6, 2, $a[9]);
parent::addRDF(parent::triplifyString($product_id, parent::getVoc() . "end-marketing-date", $date));
}
if ($a[10]) {
parent::addRDF(parent::triplifyString($product_id, parent::getVoc() . "marketing-category", $a[10]));
}
if ($a[11]) {
parent::addRDF(parent::triplifyString($product_id, parent::getVoc() . "application-number", $a[11]));
}
// create a labeller node
if ($a[12]) {
$labeller_id = parent::getRes() . md5($a[12]);
$label = addslashes($a[12]);
parent::addRDF(parent::describeIndividual($labeller_id, $label, parent::getVoc() . "Labeller") . parent::describeClass(parent::getVoc() . "Labeller", "NDC Labeller") . parent::triplify($product_id, parent::getVoc() . "labeller", $labeller_id));
}
// the next three are together
if ($a[13]) {
// MV
$substances = explode(";", $a[13]);
$strengths = explode(";", $a[14]);
$units = explode(";", $a[15]);
$l = '';
foreach ($substances as $i => $substance) {
// list the active ingredient
$ingredient_label = strtolower($substance);
$strength = '';
if (isset($strengths[$i])) {
$strength = $strengths[$i];
}
$unit = $units[$i];
$ingredient_id = parent::getRes() . md5($ingredient_label);
parent::addRDF(parent::describeIndividual($ingredient_id, $ingredient_label, parent::getVoc() . "Ingredient") . parent::describeClass(parent::getVoc() . "Ingredient", "NDC Ingredient") . parent::triplify($product_id, parent::getVoc() . "ingredient", $ingredient_id));
// describe the substance composition
$substance_label = "{$strength} {$unit} {$ingredient_label}";
$substance_id = parent::getRes() . md5($substance_label);
parent::addRDF(parent::describeIndividual($substance_id, $substance_label, parent::getVoc() . "Substance") . parent::triplifyString($substance_id, parent::getVoc() . "amount", $strength) . parent::describeClass(parent::getVoc() . "Substance", "NDC Substance"));
$unit_id = parent::getVoc() . md5($unit);
parent::addRDF(parent::describeIndividual($unit_id, $unit, parent::getVoc() . "Unit") . parent::describeClass(parent::getVoc() . "Unit", "NDC Unit") . parent::triplify($substance_id, parent::getVoc() . "amount_unit", $unit_id) . parent::triplify($product_id, parent::getVoc() . "has-part", $substance_id));
}
}
if ($a[16]) {
// MV
$b = explode(",", $a[16]);
foreach ($b as $c) {
$cat_id = parent::getVoc() . md5($c);
parent::addRDF(parent::describeIndividual($cat_id, $c, parent::getVoc() . "Pharmacological-Class") . parent::describeClass(parent::getVoc() . "Pharmacological-Class", "NDC Pharmacological Class") . parent::triplify($product_id, parent::getVoc() . "pharmacological-class", $cat_id));
}
}
parent::WriteRDFBufferToWriteFile();
}
}
function pubmed()
{
$citations = null;
$ext = substr(strrchr($this->getReadFile()->getFileName(), '.'), 1);
if ($ext = "gz") {
$citations = new SimpleXMLElement("compress.zlib://" . $this->getReadFile()->getFileName(), NULL, TRUE);
} elseif ($ext = "xml") {
$citations = new SimpleXMLElement($this->getReadFile()->getFileName(), NULL, TRUE);
}
foreach ($citations->MedlineCitation as $citation) {
$this->setCheckPoint('record');
$pmid = "" . $citation->PMID;
if (isset($this->id_list)) {
if (!isset($this->id_list[$pmid])) {
continue;
} else {
echo "processing {$pmid}" . PHP_EOL;
}
}
$pmid_uri = parent::getNamespace() . $citation->PMID;
$article = $citation->Article;
parent::addRDF(parent::describeIndividual($pmid_uri, $this->getString($article->ArticleTitle), parent::getVoc() . "PubMedRecord") . parent::describeClass(parent::getVoc() . "PubMedRecord", "PubMedRecord") . parent::triplify($pmid_uri, "rdfs:seeAlso", "http://www.ncbi.nlm.nih.gov/pubmed/{$pmid}"));
// metadata about the record
$owner = parent::getRes() . md5($citation['Owner']);
parent::addRDF(parent::describeIndividual($owner, $citation['Owner'], "foaf:Agent") . parent::triplify($pmid_uri, parent::getVoc() . "owner", $owner));
$status = parent::getRes() . md5($citation['Status']);
parent::addRDF(parent::describeIndividual($status, $citation['Status'], parent::getVoc() . "Status") . parent::describeClass(parent::getVoc() . "Status", "Status") . parent::triplify($pmid_uri, parent::getVoc() . "status", $status) . parent::triplifyString($pmid_uri, parent::getVoc() . "version", $citation['VersionID']));
$this->addDate($pmid_uri, "version-date", $citation['VersionDate']);
$this->addDate($pmid_uri, "date-created", $citation->DateCreated);
$this->addDate($pmid_uri, "date-revised", $citation->DateRevised);
$this->addDate($pmid_uri, "date-completed", $citation->DateCompleted);
if (!empty($citation->MeshHeadingList)) {
$i = 0;
foreach ($citation->MeshHeadingList->MeshHeading as $mh) {
$id = parent::getRes() . $pmid . "_mh_" . ++$i;
$did = parent::getRes() . md5($mh->DescriptorName);
parent::addRDF(parent::describeIndividual($id, $mh->DescriptorName, parent::getVoc() . "MeshHeading") . parent::describeClass(parent::getVoc() . "MeshHeading", "MeSH Heading") . parent::triplify($pmid_uri, parent::getVoc() . "mesh-heading", $id) . parent::triplifyString($id, parent::getVoc() . "descriptor-major-topic", "" . $mh->DescriptorName['MajorTopicYN']) . parent::describeIndividual($did, "" . $mh->DescriptorName, parent::getVoc() . "Mesh-Descriptor") . parent::triplify($id, parent::getVoc() . "mesh-descriptor", $did));
if (!empty($mh->QualifierName)) {
foreach ($mh->QualifierName as $qualifier_name) {
$qid = parent::getRes() . md5($qualifier_name);
parent::addRDF(parent::describeIndividual($qid, $qualifier_name, parent::getVoc() . "Mesh-Qualifier") . parent::triplify($id, parent::getVoc() . "mesh-qualifier", $qid));
}
}
}
}
if (!empty($citation->ChemicalList)) {
$i = 0;
foreach ($citation->ChemicalList->Chemical as $chemical) {
$id = parent::getRes() . $pmid . "_ch_" . ++$i;
parent::addRDF(parent::describeIndividual($id, $chemical->NameOfSubstance, parent::getVoc() . "Chemical") . parent::describeClass(parent::getVoc() . "Chemical", "Chemical") . parent::triplify($pmid_uri, parent::getVoc() . "chemical", $id));
if ($chemical->RegistryNumber != "0") {
// check if "EC"
if (substr($chemical->RegistryNumber, 0, 2) == "EC") {
$ec = substr($chemical->RegistryNumber, 3);
parent::addRDF(parent::triplify($id, parent::getVoc() . "x-ec", "ec:" . $ec));
} else {
parent::addRDF(parent::triplify($id, parent::getVoc() . "x-cas", "cas:" . $chemical->RegistryNumber));
}
}
}
}
if (!empty($citation->GeneSymbolList)) {
foreach ($citation->GeneSymbolList->GeneSymbol as $geneSymbol) {
parent::addRDF(parent::triplifyString($pmid_uri, parent::getVoc() . "gene-symbol", $geneSymbol));
}
}
if (!empty($citation->SupplMeshList)) {
foreach ($citation->SupplMeshList->SupplMeshName as $supplMeshName) {
$id = parent::getRes() . md5($supplMeshName);
parent::addRDF(parent::describeIndividual($id, $supplMeshName, parent::getVoc() . "MeshHeading") . parent::triplify($pmid_uri, parent::getVoc() . "supplemental-mesh-heading", $id));
}
}
foreach ($article->PublicationTypeList->PublicationType as $publicationType) {
$id = parent::getRes() . md5($publicationType);
$label = str_replace(" ", "-", $publicationType);
parent::addRDF(parent::triplify($pmid_uri, parent::getVoc() . "publication-type", $id) . parent::describeClass($id, $publicationType));
}
if (!empty($article->Abstract)) {
$id = parent::getRes() . $pmid . "_ABSTRACT";
$label = "Abstract for PMID:{$pmid}";
$abstract = $article->Abstract;
parent::addRDF(parent::describeIndividual($id, $label, parent::getVoc() . "Article-Abstract") . parent::describeClass(parent::getVoc() . "Article-Abstract", "Article Abstract") . parent::triplify($pmid_uri, "dc:abstract", $id) . parent::triplifyString($id, parent::getVoc() . "copyright", $abstract->CopyrightInformation));
$section = 0;
$abstractText = "";
foreach ($abstract->AbstractText as $text) {
$abstractText .= " " . $text;
if (!empty($text['Label']) && $text['Label'] !== "UNLABELLED") {
$section_id = parent::getRes() . $pmid . "_ABSTRACT_SECTION_" . ++$section;
parent::addRDF(parent::triplify($id, parent::getVoc() . "section", $section_id) . parent::triplifyString($section_id, parent::getVoc() . "order", $section) . parent::triplifyString($section_id, parent::getVoc() . "nlm-section-type", $text['NlmCategory']) . parent::triplifyString($section_id, parent::getVoc() . "label", $text['Label']) . parent::triplifyString($section_id, parent::getVoc() . "text", $text));
}
}
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "abstract-text", $abstractText));
}
if (!empty($citation->OtherAbstract)) {
$i = 0;
foreach ($citation->OtherAbstract as $ab) {
$id = parent::getRes() . $pmid . "_oa_" . ++$i;
parent::addRDF(parent::describeIndividual($id, "", parent::getVoc() . "Article-Abstract") . parent::describeClass(parent::getVoc() . "Article-Abstract", "Article Abstract") . parent::triplify($pmid_uri, "dc:abstract", $id));
$abstractText = "";
foreach ($ab->AbstractText as $text) {
$abstractText .= " " . $text;
if (!empty($text['Label']) && $text['Label'] !== "UNLABELLED") {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "abstract_" . strtolower($text['Category']), $text));
}
}
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "abstract-text", $abstractText));
}
}
$author_types = array("Investigator", "Author", "PersonalNameSubject");
foreach ($author_types as $author_type) {
$listname = $author_type . "List";
if (!empty($article->{$listname}->{$author_type})) {
$i = 0;
foreach ($article->{$listname}->{$author_type} as $author) {
$id = parent::getRes() . $pmid . "_AUTHOR_" . ++$i;
$author_label = $author->LastName . ($author->Initials ? ", " . $author->Initials : "");
parent::addRDF(parent::describeIndividual($id, $author_label, parent::getVoc() . $author_type) . parent::describeClass(parent::getVoc() . $author_type, $author_type) . parent::triplifyString($id, parent::getVoc() . "list-position", $i) . parent::triplify($pmid_uri, parent::getVoc() . strtolower($author_type), $id) . parent::triplifyString($id, parent::getVoc() . "last-name", $author->LastName) . parent::triplifyString($id, parent::getVoc() . "fore-name", $author->ForeName) . parent::triplifyString($id, parent::getVoc() . "initials", $author->Initials) . parent::triplifyString($id, parent::getVoc() . "collective-name", $author->CollectiveName) . parent::triplifyString($id, parent::getVoc() . "suffix", $author->Suffix));
if ($author->Affiliation) {
$affilitation = parent::getRes() . md5($author->Affilitation);
parent::addRDF(parent::describeIndividual($affilitation, $author->Affilitation, parent::getVoc() . "Organization") . parent::describeClass(parent::getVoc() . "Organization", "Organization") . parent::triplifyString($id, parent::getVoc() . "affiliation", $affilitation));
}
foreach ($author->NameID as $authorNameId) {
if (!empty($authorNameId)) {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "name-id", $author_name_id));
}
}
}
}
}
if (!empty($article->ArticleDate)) {
$this->addDate($pmid_uri, "article-date", $article->ArticleDate);
}
foreach ($article->Language as $language) {
parent::addRDF(parent::triplifyString($pmid_uri, "dc:language", $language));
}
if (!empty($citation->KeywordList)) {
foreach ($citation->KeywordList->Keyword as $keyword) {
parent::addRDF(parent::triplifyString($pmid_uri, parent::getVoc() . "keyword", $keyword));
}
}
if (!empty($citation->otherID)) {
// untested
foreach ($citation->OtherID as $otherID) {
if (!empty($otherID)) {
parent::addRDF(parent::triplifyString($pmid_uri, parent::getVoc() . "other-id", $other_id) . parent::triplifyString($pmid_uri, parent::getVoc() . "other-id-source", $otherID['Source']));
if (strstr($other_id, "PMC")) {
parent::addRDF(parent::triplify($pmid_uri, parent::getVoc() . "x-pmc", "pmc:" . $other_id));
}
}
}
}
if (!empty($article->DataBankList)) {
foreach ($article->DataBankList->DataBank as $dataBank) {
parent::addRDF(parent::triplifyString($pmid_uri, parent::getVoc() . "databank", $dataBank->DataBankName));
if ($dataBank->AccessionNumberList !== NULL) {
foreach ($dataBank->AccessionNumberList->AccessionNumber as $acc) {
parent::addRDF(parent::triplifyString($pmid_uri, parent::getVoc() . "x-" . strtolower($dataBank->dataBankName), $acc));
}
}
}
}
if (!empty($article->GrantList)) {
$i = 0;
foreach ($article->GrantList->Grant as $grant) {
$id = parent::getRes() . $pmid . "_GRANT_" . ++$i;
$grant_label = "Grant " . $grant->GrantID . " for " . parent::getNamespace() . $pmid;
parent::addRDF(parent::describeIndividual($id, $grant_label, parent::getVoc() . "Grant") . parent::describeClass(parent::getVoc() . "Grant", "Grant") . parent::triplify($pmid_uri, parent::getVoc() . "grant", $id) . parent::triplifyString($id, parent::getVoc() . "grant-identifier", $grant->GrantID) . parent::triplifyString($id, parent::getVoc() . "grant-acronym", $grant->Acronym) . parent::triplifyString($id, parent::getVoc() . "grant-agency", $grant->Agency) . parent::triplifyString($id, parent::getVoc() . "grant-country", $grant->Country));
}
}
if (!empty($citation->NumberOfReferences)) {
parent::addRDF(parent::triplifyString($pmid_uri, parent::getVoc() . "number-of-references", $citation->NumberOfReferences));
}
if (!empty($article->VernacularTitle)) {
parent::addRDF(parent::triplifyString($pmid_uri, parent::getVoc() . "vernacular-title", $article->VernacularTitle));
}
foreach ($citation->CitationSubset as $citationSubset) {
if (!empty($citationSubset)) {
parent::addRDF(parent::triplifyString($pmid_uri, parent::getVoc() . "citation-subset", $citationSubset));
}
}
if (!empty($citation->commentsCorrectionsList)) {
$i = 0;
foreach ($commentsCorrectionsList->CommentsCorrections as $commentCorrection) {
$id = parent::getRes() . $pmid . "_COMMENT_CORRECTION_" . ++$i;
$ccRefType = $commentCorrection['RefType'];
$ccPmid = $commentCorrection->PMID;
//optional
$ccNote = $commentCorrection->Note;
//optional
$cc_label = "Comment or correction ." . $ccNumber . " for " . parent::getNamespace() . $pmid;
parent::addRDF(parent::describeIndividual($id, $cc_label, parent::getVoc() . "CommentCorrection") . parent::describeClass(parent::getVoc() . "CommentCorrection", "CommentCorrection") . parent::triplify($pmid_uri, parent::getVoc() . "comment-correction", $id) . parent::triplify($id, "rdf:type", parent::getVoc() . $ccRefType) . parent::triplifyString($id, parent::getVoc() . "ref-source", $ref_source) . parent::triplifyString($id, parent::getVoc() . "note", $cc_note));
}
}
if (!empty($citation->generalNote)) {
parent::addRDF(parent::triplifyString($pmid_uri, parent::getVoc() . "general-note", $general_note));
}
foreach ($citation->SpaceFlightMission as $spaceFlightMission) {
if (!empty($spaceFlightMission)) {
parent::addRDF(parent::triplifyString($pmid_uri, parent::getVoc() . "space-flight-mission" . $space_flight_mission));
}
}
$journal = $article->Journal;
$journalId = parent::getRes() . $pmid . "_JOURNAL";
$journal_label = "Journal for " . parent::getNamespace() . $pmid;
parent::addRDF(parent::describeIndividual($journalId, $journal_label, parent::getVoc() . "Journal") . parent::describeClass(parent::getVoc() . "Journal", "Journal") . parent::triplify($pmid_uri, parent::getVoc() . "journal", $journalId) . parent::triplify($journalId, parent::getVoc() . "x-issn", "issn:" . $journal->ISSN) . parent::triplifyString($journalId, parent::getVoc() . "journal-nlm-identifier", $citation->MedLineJournalInfo->NlmUniqueID) . parent::triplifyString($journalId, parent::getVoc() . "journal-title", $journal->Title) . parent::triplifyString($journalId, parent::getVoc() . "journal-abbreviation", $journal->ISOAbbreviation) . parent::triplifyString($journalId, parent::getVoc() . "volume", $journal->JournalIssue->Volume) . parent::triplifyString($journalId, parent::getVoc() . "issue", $journal->JournalIssue->Issue) . parent::triplifyString($journalId, parent::getVoc() . "pages", "" . $article->Pagination->MedlinePgn));
$journalPubDate = $journal->JournalIssue->PubDate;
if (!empty($journalPubDate)) {
$journalYear = $journalPubDate->Year;
$journalMonth = trim($journalPubDate->Month);
//optional
if ($journalMonth and !is_numeric($journalMonth[0])) {
$mo = array("jan", "feb", "mar", "apr", "may", "jun", "jul", "aug", "sep", "oct", "nov", "dec");
$journalMonth = str_pad(array_search(strtolower($journalMonth), $mo) + 1, 2, "0", STR_PAD_LEFT);
}
$journalDay = trim($journalPubDate->Day);
//optional
if ($journalDay) {
$journalDay = str_pad($journalDay, 2, "0", STR_PAD_LEFT);
}
parent::addRDF(parent::triplifyString($journalId, parent::getVoc() . "publication-year", $journalYear) . parent::triplifyString($journalId, parent::getVoc() . "publication-month", $journalMonth) . parent::triplifyString($journalId, parent::getVoc() . "publication-day", $journalDay) . parent::triplifyString($journalId, parent::getVoc() . "publication-season", $journalPubDate->Season) . parent::triplifyString($journalId, parent::getVoc() . "publication-date", $journalPubDate->MedlineDate));
if (!empty($journalYear) and !empty($journalMonth) and !empty($journalDay)) {
parent::addRDF(parent::triplifyString($journalId, parent::getVoc() . "publication-date", "{$journalYear}-{$journalMonth}-{$journalDay}", "xsd:date"));
}
}
foreach ($citation->Article->ELocation as $eLocation) {
if (!empty($eLocation)) {
parent::addRDF(parent::triplifyString($pmid_uri, parent::getVoc() . "elocation", $eLocation));
}
}
$this->writeRDFBufferToWriteFile();
//break;
}
}
function models()
{
$tax_ids = array("Caenorhabditis elegans" => "6239", "Mus musculus" => "10090", "Saccharomyces cerevisiae" => "4932", "Drosophila melanogaster" => "7227", "Podospora anserina" => "5145", "Mesocricetus auratus" => "10036", "Schizosaccharomyces pombe" => "4896", "Danio rerio" => "7955");
$h = explode(",", parent::getReadFile()->read());
$expected_columns = 8;
if (($n = count($h)) != $expected_columns) {
trigger_error("Found {$n} columns in gene file - expecting {$expected_columns}!", E_USER_WARNING);
return false;
}
/*
[0] GenAge ID
[1] symbol
[2] name
[3] organism
[4] entrez gene id
[5] avg lifespan change (max obsv)
[6] lifespan effect
[7] longevity influence
*/
while ($l = parent::getReadFile()->read(200000)) {
$data = str_getcsv($l);
$genage = str_pad($data[0], 4, "0", STR_PAD_LEFT);
$gene_symbol = $data[1];
$name = $data[2];
$organism = $data[3];
$ncbi_gene_id = $data[4];
$max_percent_obsv_avg_lifespan_change = $data[5];
$lifespan_effect = $data[6];
$longevity_influence = $data[7];
$genage_id = parent::getNamespace() . $genage;
parent::addRDF(parent::describeIndividual($genage_id, $name, parent::getVoc() . "Aging-Related-Gene") . parent::describeClass(parent::getVoc() . "Aging-Related-Gene", "Aging Related Gene"));
parent::addRDF(parent::triplifyString($genage_id, parent::getVoc() . "gene-symbol", parent::safeLiteral($gene_symbol)));
parent::addRDF(parent::triplify($genage_id, parent::getVoc() . "taxon", "ncbitaxon:" . $tax_ids[$organism]));
if ($ncbi_gene_id !== "") {
parent::addRDF(parent::triplify($genage_id, parent::getVoc() . "x-ncbigene", "ncbigene:" . $ncbi_gene_id));
}
if ($max_percent_obsv_avg_lifespan_change !== "") {
parent::addRDF(parent::triplifyString($genage_id, parent::getVoc() . "maximum-percent-observed-average-lifespan-change", parent::safeLiteral($max_percent_obsv_avg_lifespan_change)));
}
if ($lifespan_effect == "Increase and Decrease") {
parent::addRDF(parent::triplifyString($genage_id, parent::getVoc() . "lifespan-effect", "increase") . parent::triplifyString($genage_id, parent::getVoc() . "lifespan-effect", "decrease"));
} else {
parent::addRDF(parent::triplifyString($genage_id, parent::getVoc() . "lifespan-effect", strtolower($lifespan_effect)));
}
parent::addRDF(parent::triplifyString($genage_id, parent::getVoc() . "longevity-influence", strtolower($longevity_influence)));
parent::WriteRDFBufferToWriteFile();
}
}
function gene_manipulations()
{
$h = explode(",", parent::getReadFile()->read());
$expected_columns = 5;
if (($n = count($h)) != $expected_columns) {
trigger_error("Found {$n} columns in gene file - expecting {$expected_columns}!", E_USER_WARNING);
return false;
}
while ($l = parent::getReadFile()->read(200000)) {
$data = str_getcsv($l);
$gendr = $data[0];
$gene_symbol = $data[1];
$species_name = $data[2];
$geneid = $data[3];
$gene_name = $data[4];
// $references = $data[5];
$gendr_id = parent::getNamespace() . $gendr;
$gendr_label = $gene_name . " (" . $gene_symbol . ")";
$association_id = parent::getRes() . md5($gendr . $geneid . "_association");
$association_label = "Association between " . $gene_symbol . " and variation in life span extension induced by dietary restriction";
parent::addRDF(parent::describeIndividual($gendr_id, $gendr_label, parent::getVoc() . "Dietary-Restriction-Life-Extension-Related-Gene") . parent::describeClass(parent::getVoc() . "Dietary-Restriction-Life-Extension-Related-Gene", "Dietery Restriction Life Extension Related Gene") . parent::triplify($gendr_id, parent::getVoc() . "x-ncbigene", "ncbigene:" . $geneid) . parent::triplifyString($gendr_id, parent::getVoc() . "gene-name", $gene_name) . parent::triplifyString($gendr_id, parent::getVoc() . "gene-symbol", $gene_symbol) . parent::describeIndividual($association_id, $association_label, parent::getVoc() . "Gene-Phenotype-Association") . parent::describeClass(parent::getVoc() . "Gene-Phenotype-Association", "Gene Phenotype Association") . parent::triplify($association_id, parent::getVoc() . "gene", $gendr_id) . parent::triplify($association_id, parent::getVoc() . "phenotype", parent::getVoc() . "Diet-Induced-Life-Span-Variant"));
if ($species_name == "Caenorhabditis elegans") {
parent::addRDF(parent::triplify($association_id, parent::getVoc() . "phenotype", "wormbase:WBPhenotype:0001837") . parent::triplify($association_id, parent::getVoc() . "taxon", "taxon:6239"));
} else {
if ($species_name == "Saccharomyces cerevisiae") {
parent::addRDF(parent::triplify($association_id, parent::getVoc() . "taxon", "taxon:4932"));
} else {
if ($species_name == "Schizosaccharomyces pombe") {
parent::addRDF(parent::triplify($association_id, parent::getVoc() . "taxon", "taxon:9896"));
} else {
if ($species_name == "Drosophila melanogaster") {
parent::addRDF(parent::triplify($association_id, parent::getVoc() . "taxon", "taxon:7227"));
} else {
if ($species_name == "Mus musculus") {
parent::addRDF(parent::triplify($association_id, parent::getVoc() . "taxon", "taxon:10090"));
}
}
}
}
}
if (!empty($references)) {
$split_refs = explode(",", $references);
foreach ($split_refs as $ref) {
parent::addRDF(parent::triplify($gendr_id, parent::getVoc() . "article", "pmid:" . $ref) . parent::triplify($association_id, parent::getVoc() . "article", "pmid:" . $ref));
}
}
parent::writeRDFBufferToWriteFile();
}
//while
}
/**
* add an RDF representation of the incoming param to the model.
* @$desc_record_arr is an assoc array with the contents of one qualifier record
*/
private function makeDescriptorRecord($desc_record_arr)
{
//get the UI of the descriptor record
$dr_ui = $desc_record_arr["UI"][0];
$dr_res = $this->getNamespace() . $dr_ui;
$dr_label = $desc_record_arr['MH'][0];
parent::AddRDF(parent::describeIndividual($dr_res, $dr_label, $this->getVoc() . "Descriptor", $dr_label) . parent::describeClass($this->getVoc() . "Descriptor", "MeSH Descriptor"));
//now get the descriptor_data_elements
$qde = $this->getDescriptorDataElements();
//iterate over the properties
foreach ($desc_record_arr as $k => $v) {
if (array_key_exists($k, $qde)) {
if ($k == "AN") {
foreach ($v as $kv => $vv) {
//explode by semicolon
$vvrar = explode(";", $vv);
foreach ($vvrar as $anAn) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde["AN"], $anAn) . parent::describeProperty($this->getVoc() . $qde["AN"], "Relationship between a descriptor and its annotation"));
}
//foreach
}
//foreach
}
//if
//add allowable topical qualifiers
if ($k == "AQ") {
//$x = $this->getDescriptorDataElements();
foreach ($v as $kv => $vv) {
$vvrar = explode(" ", $vv);
foreach ($vvrar as $aq) {
$aq_res = $this->getRes() . $aq;
parent::AddRDF(parent::triplify($aq_res, "rdf:type", $this->getVoc() . "allowable-topical-qualifier") . parent::describeClass($this->getVoc() . "allowable-topical-qualifier", "allowable topical qualifier: " . $qde['AQ']));
parent::AddRDF(parent::triplify($dr_res, $this->getVoc() . $qde['AQ'], $aq_res) . parent::describeProperty($this->getVoc() . $qde['AQ'], "Relationship between a descriptor and its allowable topical qualifiers"));
}
//foreach
}
//foreach
}
//if
//add CATALOGING SUBHEADINGS LIST NAME
if ($k == "CATSH") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['CATSH'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['CATSH'], "Relationship between a descriptor and its cataloging subheadings list name"));
}
}
//if
if ($k == "CX") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['CX'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['CATSH'], "Relationship between a descriptor and xrefs"));
}
}
//if
//add date of entry
if ($k == "DA") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['DA'], $this->formatDate($vv), "xsd:date") . parent::describeProperty($this->getVoc() . $qde['DA'], "Relationship between a descriptor and its date of entry"));
}
}
//if
//descriptor class
if ($k == "DC") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['DC'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['DC'], "Relationship between a descriptor and its descriptor class"));
}
}
//if
//descriptor entry version
if ($k == "DE") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['DE'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['DE'], "Relationship between a descriptor record and its entry version"));
}
}
//if
//descriptor sort version
if ($k == "DS") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['DS'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['DS'], "Relationship between a descriptor record and its sort version"));
}
}
//if
//date major descriptor established
if ($k == "DX") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['DX'], $this->formatDate($vv), "xsd:date") . parent::describeProperty($this->getVoc() . $qde['DX'], "Relationship between a descriptor and its date of major descriptor established"));
}
}
//if
if ($k == "EC") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['EC'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['EC'], "Relationship between a descriptor and its entry combination"));
}
}
if ($k == "PRINT ENTRY") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['PRINT ENTRY'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['PRINT ENTRY'], "Relationship between a descriptor and its print entry term"));
}
}
if ($k == "ENTRY") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['ENTRY'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['ENTRY'], "Relationship between a descriptor and its entry term"));
}
}
if ($k == "FX") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['FX'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['FX'], "Relationship between a descriptor and its forward cross reference"));
}
}
if ($k == "GM") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['GM'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['GM'], "Relationship between a descriptor and its grateful med note"));
}
}
if ($k == "HN") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['HN'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['HN'], "Relationship between a descriptor record and its history note"));
}
}
if ($k == "MED") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['MED'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['MED'], "Relationship between a descriptor and its backfile postings"));
}
}
if ($k == "M94") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['M94'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['M94'], "Relationship between a descriptor and its backfile postings"));
}
}
if ($k == "M90") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['M90'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['M90'], "Relationship between a descriptor and its backfile postings"));
}
}
if ($k == "M85") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['M85'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['M85'], "Relationship between a descriptor record and its backfile postings"));
}
}
if ($k == "M80") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['M80'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['M80'], "Relationship between a descriptor record and its backfile postings"));
}
}
if ($k == "M75") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['M75'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['M75'], "Relationship between a descriptor record and its backfile postings"));
}
}
if ($k == "M66") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['M66'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['M66'], "Relationship between a descriptor record and its backfile postings"));
}
}
if ($k == "MH_TH") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['MH_TH'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['MH_TH'], "Relationship between a descriptor record and its MeSH Heading thesaurus id"));
}
}
if ($k == "MH") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['MH'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['MH'], "Relationship between a descriptor record and its MeSH Heading"));
}
}
if ($k == "MN") {
foreach ($v as $kv => $vv) {
$vid = parent::getNamespace() . $vv;
$vlabel = utf8_encode(htmlspecialchars($vv));
parent::AddRDF(parent::describeIndividual($vid, $dr_label, parent::getVoc() . "Tree-Entry", $vlabel) . parent::triplify($dr_res, $this->getVoc() . $qde['MN'], $vid) . parent::describeProperty($this->getVoc() . $qde['MN'], "Relationship between a descriptor record and its MeSH Tree Number"));
if (FALSE !== ($pos = strrpos($vv, "."))) {
$pid = parent::getNamespace() . substr($vv, 0, $pos);
parent::addRDF(parent::triplify($vid, "rdfs:subClassOf", $pid));
}
}
}
if ($k == "MR") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['MR'], $this->formatDate($vv), "xsd:date") . parent::describeProperty($this->getVoc() . $qde['MR'], "Relationship between a descriptor record and its major revision date"));
}
}
if ($k == "MS") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['MS'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['MS'], "Relationship between a descriptor record and its MeSH scope note"));
}
}
if ($k == "N1") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['N1'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['N1'], "Relationship between a descriptor record and its CAS 1 name"));
}
}
if ($k == "OL") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['OL'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['OL'], "Relationship between a descriptor record and its online note"));
}
}
if ($k == "PA") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['PA'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['PA'], "Relationship between a descriptor record and its pharmacological action"));
}
}
if ($k == "PI") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['PI'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['PI'], "Relationship between a descriptor record and its previous indexing"));
}
}
if ($k == "PM") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['PM'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['PM'], "Relationship between a descriptor record and its public mesh note"));
}
}
if ($k == "PX") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['PX'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['PX'], "Relationship between a descriptor record and its pre explosion"));
}
}
if ($k == "RECTYPE") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['RECTYPE'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['RECTYPE'], "Relationship between a descriptor record and its record type"));
}
}
if ($k == "RH") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['RH'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['RH'], "Relationship between a descriptor record and its running head, in relation to mesh tree structures"));
}
}
if ($k == "RN") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['RN'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['RN'], "Relationship between a descriptor record and its CAS registry"));
}
}
if ($k == "RR") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($dr_res, $this->getVoc() . $qde['RR'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['RR'], "Relationship between a descriptor record and its registry number"));
}
}
if ($k == "ST") {
foreach ($v as $kv => $vv) {
$vid = parent::getNamespace() . $vv;
$pid = parent::getNamespace() . substr($vv, 0, strrpos($vv, ".") - 1);
$vlabel = utf8_encode(htmlspecialchars($vv));
parent::AddRDF(parent::describeIndividual($vid, $vlabel, parent::getVoc() . "Semantic-Type", $vlabel) . parent::triplify($dr_res, $this->getVoc() . $qde['ST'], $vid) . parent::describeProperty($this->getVoc() . $qde['ST'], "Relationship between a descriptor record and its semantic type"));
}
}
} else {
trigger_error("Please add key to descriptor record map: " . $k . PHP_EOL, E_USER_ERROR);
}
$this->WriteRDFBufferToWriteFile();
}
//foreach
$this->WriteRDFBufferToWriteFile();
}
function parseEntry($lfile)
{
$fp = fopen($lfile, "r");
while ($l = fgets($fp, 100000)) {
$k_t = trim(substr($l, 0, 12));
$v = trim(substr($l, 12));
if (!$k_t and $v == '') {
continue;
}
// set the key to the current key if not empty, else keep using what was there before
if (!isset($k)) {
$k = $k_t;
} else {
if (!empty($k_t)) {
$k = $k_t;
}
}
if ($k == "///" or $k == "ENTRY1") {
break;
}
if ($k == "ENTRY") {
$a = explode(" ", $v, 2);
$e['id'] = str_replace(array("EC ", " "), "", $a[0]);
if (isset($this->org)) {
$e['id'] = $this->org . "_" . $e['id'];
}
$e['type'] = trim(str_replace(array("Complete ", "Pathway Module"), array("", "Pathway Module"), $a[1]));
$e['type_label'] = str_replace(" ", "-", $e['type']);
$uri = parent::getNamespace() . $e['id'];
continue;
}
// key with value
if (in_array($k, array("NAME", "DESCRIPTION", "DEFINITION", "EQUATION", "COMMENT"))) {
if ($k == "NAME") {
parent::addRDF(parent::describeIndividual($uri, $v, parent::getVoc() . $e['type']) . parent::describeClass(parent::getVoc() . $e['type'], $e['type_label']) . parent::triplify($uri, "rdfs:seeAlso", "http://www.kegg.jp/dbget-bin/www_bget?" . $e['id']));
if ($e['type'] == 'Genome') {
$a = explode(",", $v);
parent::addRDF(parent::triplify($uri, "owl:sameAs", "kegg:" . $a[0]));
}
} else {
if ($k == "DESCRIPTION") {
parent::addRDF(parent::triplifyString($uri, "dc:description", $v));
} else {
if ($k == "DEFINITION" and $e['type'] == "KO") {
preg_match("/\\[([^\\]]+)\\]/", $v, $m);
if (isset($m[1])) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . "x-ec", $m[1]));
}
} else {
if ($k == "COMMENT") {
preg_match("/ICD-O: ([^,]+),/", $v, $m);
if (isset($m[1])) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . "x-icdo", "icdo:" . $m[1]));
continue;
}
} else {
parent::addRDF(parent::triplifyString($uri, parent::getVoc() . strtolower($k), $v));
}
}
}
}
continue;
}
if ($k == "RPAIR" and $e['type'] == "Reaction") {
$list = explode(" ", $v);
$id = parent::getRes() . $e['id'] . "." . $list[2] . "." . $list[3];
$rc = '';
if (isset($list[4])) {
$rc = "kegg:" . substr($list[4], 4, -1);
}
parent::addRDF(parent::describeIndividual($id, $e['id'] . " " . $v, parent::getVoc() . "RPair-Role") . parent::describeClass(parent::getVoc() . "RPair-Role", "RPair Role") . parent::triplify($id, parent::getVoc() . "rpair", "kegg:" . $list[0]) . parent::triplifyString($id, parent::getVoc() . "role", $list[3]) . ($rc != '' ? parent::triplify($id, parent::getVoc() . "reaction-center", $rc) : '') . parent::triplify($uri, parent::getVoc() . "rpair-role", $id));
continue;
}
// list of entries
if (in_array($k, array("ENZYME", "RPAIR", "RELATEDPAIR")) or in_array($e['type'], array("Compound", "RClass", "RPair")) and $k == "REACTION") {
$list = explode(" ", $v);
foreach ($list as $id) {
if (!$id) {
continue;
}
parent::addRDF(parent::triplify($uri, parent::getVoc() . strtolower($k), "kegg:{$id}"));
}
continue;
}
// key with semi-colon separated values
if (in_array($k, array("CLASS", "CATEGORY", "KEYWORDS", "CHROMOSOME", "ANNOTATION", "ACTIVITY", "TYPE"))) {
$a = explode(";", $v);
foreach ($a as $c) {
parent::addRDF(parent::triplifyString($uri, parent::getVoc() . strtolower($k), trim($c)));
}
continue;
}
// kegg seems to make a prefix mistake with the pathway identifiers...
if ($k == "PATHWAY") {
$a = explode(" ", $v, 2);
preg_match("/[a-z]+([0-9]{5})/", $a[0], $m);
if (isset($m[1])) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . strtolower($k), "kegg:map" . $m[1]));
} else {
echo "pathway problem: " . $v . PHP_EOL;
}
continue;
}
// multi-line header with key-value pair
if (in_array($k, array("PATHWAY_MAP", "STR_MAP", "MODULE", "DISEASE", "KO_PATHWAY", "COMPOUND"))) {
// PATHWAY_MAP map00010 Glycolysis / Gluconeogenesis
$a = explode(" ", $v, 2);
$mid = $a[0];
if (strpos($a[0], '(') !== FALSE) {
$mid = substr($a[0], 0, strpos($a[0], '('));
}
if (isset($this->org) and $k == "MODULE") {
$mid = substr($mid, strpos($v, "_") + 1);
}
parent::addRDF(parent::triplify($uri, parent::getVoc() . strtolower($k), "kegg:" . $mid));
continue;
}
// REACTION parsing
if (preg_match("/\\[RN:([^\\]]+)]/", $v, $m) != FALSE) {
$list = explode(" ", $m[1]);
foreach ($list as $item) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . strtolower($k), "kegg:" . $item));
}
continue;
}
if ($k == "DRUG") {
preg_match("/\\[DR:([^\\]]+)]/", $v, $m);
if (isset($m[1])) {
$list = explode(" ", $m[1]);
foreach ($list as $item) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . strtolower($k), "kegg:" . $item));
}
continue;
}
}
if ($k == "TAXONOMY") {
parent::addRDF(parent::triplify($uri, parent::getVoc() . strtolower($k), "kegg:" . str_replace("TAX", "taxonomy", $v)));
continue;
}
// a list of objects to parse out that are defined within square brackets
if (in_array($k, array("SOURCE", "COMPONENT"))) {
preg_match_all("/\\[([^\\]]+)\\]/", $v, $m);
if (isset($m[1])) {
foreach ($m[1] as $id) {
$myid = str_replace(array("TAX", "CPD", "DR"), array("taxonomy", "kegg", "kegg"), $id);
parent::addRDF(parent::triplify($uri, parent::getVoc() . strtolower($k), $myid));
}
continue;
}
}
// multi-line header with multi-key single value pair
if (in_array($k, array("ORTHOLOGY", "REACTION"))) {
// K00844,K12407,K00845 hexokinase/glucokinase [EC:2.7.1.1 2.7.1.2] [RN:R01786]
// R01786,R02189,R09085 C00267 -> C00668
$a = explode(" ", $v, 2);
$ids = explode(",", $a[0]);
if ($k == "REACTION" and $ids[0][0] != "R") {
echo "unable to parse {$k}" . PHP_EOL;
continue;
}
if (!isset($a[1])) {
if ($e['type'] == "Reaction") {
parent::addRDF(parent::triplify($uri, parent::getVoc() . "orthology", "kegg:" . trim($a[0])));
continue;
}
echo "parse error: " . $k . " " . $v . PHP_EOL;
continue;
}
$str = $a[1];
foreach ($ids as $id) {
$o = '';
$o['id'] = $id;
$o['label'] = $str;
$o['type'] = strtolower($k);
parent::addRDF(parent::triplify($uri, parent::getVoc() . strtolower($k), "kegg:{$id}"));
}
continue;
}
if ($k == "DBLINKS") {
// DBLINKS GO: 0006096 0006094
$a = explode(": ", $v, 2);
$ns = str_replace(array("ncbi-geneid", "ncbi-gi", "rn", "pubchem", "pdb-ccd", "icd-10", "um-bbd", "iubmb enzyme nomenclature", "explorenz - the enzyme database", "expasy - enzyme nomenclature database", "umbbd (biocatalysis/biodegradation database)", "brenda, the enzyme database"), array("ncbigene", "gi", "kegg", "pubchem.compound", "ccd", "icd10", "umbbd", "ec", "ec", "ec", "ec", "ec"), strtolower($a[0]));
$ids = explode(" ", $a[1]);
foreach ($ids as $id) {
if (!$id) {
continue;
}
parent::addRDF(parent::triplify($uri, parent::getVoc() . "x-{$ns}", "{$ns}:{$id}"));
}
continue;
}
if ($k == "REMARK") {
preg_match("/Same as: ([A-Z0-9]+)/", $v, $m);
if (isset($m[1])) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . "same-as", "kegg:" . $m[1]));
continue;
}
preg_match("/ATC code: (.*)/", $v, $m);
if (isset($m[1])) {
$list = explode(" ", $m[1]);
foreach ($list as $item) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . "x-atc", "atc:" . $item));
}
continue;
}
preg_match("/Therapeutic category: (.*)/", $v, $m);
if (isset($m[1])) {
$list = explode(" ", $m[1]);
foreach ($list as $item) {
parent::addRDF(parent::triplifyString($uri, parent::getVoc() . "therapeutic-category", $item));
}
continue;
}
preg_match("/Drug group: (.*)/", $v, $m);
if (isset($m[1])) {
$list = explode(" ", $m[1]);
foreach ($list as $item) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . "drug-group", "kegg:" . $item));
}
continue;
}
}
if ($k == "PRODUCT" or $k == "SUBSTRATE") {
preg_match("/([a-z0-9]{8}-[a-z0-9]{4}-[a-z0-9]{4}-[a-z0-9]{4}-[a-z0-9]{12})/", $v, $m);
if (isset($m[1])) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . "x-dailymed", "dailymed:" . $m[1]) . parent::triplifyString("dailymed:" . $m[1], "rdfs:label", $v));
continue;
}
preg_match("/\\[CPD:([^\\]]+)\\]/", $v, $m);
if (isset($m[1])) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . strtolower($k), "kegg:" . $m[1]));
continue;
}
}
if ($k == "STATISTICS") {
$a = explode(": ", $v);
parent::addRDF(parent::triplifyString($uri, parent::getVoc() . str_replace(" ", "-", strtolower($a[0])), $a[1]));
continue;
}
if ($k == "ORGANISM") {
$a = explode(" ", $v);
parent::addRDF(parent::triplify($uri, parent::getVoc() . "organism", "kegg:" . $a[0]));
continue;
}
if ($k == "REFERENCE") {
if (!isset($ref)) {
$ref = 1;
} else {
if (!isset($e['reference'][$ref]['title'])) {
continue;
}
// this is a bug where the reference declaration is split onto two lines
$ref++;
}
if (strstr($v, "PMID")) {
// PMID:11529849 (marker)
preg_match("/(PMID:[0-9]+) /", $v, $m);
if (isset($m[1])) {
$e['reference'][$ref]['pubmed'] = $m[1];
}
}
continue;
}
if ($k == "AUTHORS") {
$e['reference'][$ref]['authors'] = $v;
continue;
}
if ($k == "TITLE") {
$e['reference'][$ref]['title'] = $v;
continue;
}
if ($k == "JOURNAL") {
$e['reference'][$ref]['journal'] = $v;
continue;
}
if ($e['type'] == "Disease" and ($k == "GENE" or $k == "MARKER")) {
// BCR-ABL (translocation) [HSA:613 25] [KO:K08878 K06619]
preg_match_all("/ \\[([^\\]]+)\\]/", $v, $m);
if (isset($m[1])) {
foreach ($m[1] as $idlist) {
$a = explode(":", $idlist);
$ns = $a[0];
$b = explode(" ", $a[1]);
foreach ($b as $id) {
if ($ns == "KO") {
$rel = "ko-" . strtolower($k);
$gene = $id;
} else {
$rel = strtolower($k);
$gene = $ns . "_" . $id;
}
parent::addRDF(parent::triplify($uri, parent::getVoc() . $rel, "kegg:{$gene}"));
}
}
} else {
echo $v;
}
continue;
}
if ($k == "GENES") {
// ATH: AT1G32780 AT1G64710 AT1G77120(ADH1) AT5G24760
$a = explode(": ", $v);
$org = $a[0];
$b = explode(" ", $a[1]);
foreach ($b as $id) {
$c = explode("(", $id);
$gene = parent::getNamespace() . $org . "_" . $c[0];
parent::addRDF(parent::triplify($uri, parent::getVoc() . "gene", $gene));
}
//echo parent::getRDF();exit;
continue;
}
if ($k == "DRUG_TARGET") {
// Afatinib: D09724 D09733
$s = substr($v, strpos($v, ":") + 2);
$list = explode(" ", $s);
foreach ($list as $item) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . "drug-target", "kegg:{$item}"));
}
continue;
}
if ($k == "STRUCTURE") {
$list = explode(" ", $v);
foreach ($list as $item) {
if (trim($item) == '') {
continue;
}
parent::addRDF(parent::triplify($uri, parent::getVoc() . "x-pdb", "pdb:{$item}"));
}
continue;
}
if ($k == "MOTIF") {
$list = explode(" ", $v);
foreach ($list as $item) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . "x-pfam", "pfam:{$item}"));
}
continue;
}
if (in_array($k, array("INTERACTION", "METABOLISM", "TARGET"))) {
// dopamine D2-receptor antagonist [HSA:1813] [KO:K04145]
$id = parent::getRes() . md5($uri . $v);
$type = ucfirst(strtolower($k));
if (in_array($k, array("INTERACTION", "METABOLISM"))) {
$a = explode(":", $v, 2);
$modifier = $a[0];
} else {
$modifier = '';
$s = substr($v, 0, strpos($v, "[") + 1);
// dopamine D2-receptor antagonist [
preg_match("/ ([a-z]+) \\[/", $s, $m);
if (isset($m[1])) {
$modifier = $m[1];
}
}
parent::addRDF(parent::describeIndividual($id, $v, parent::getVoc() . $type) . parent::describeClass(parent::getVoc() . $type, $type) . parent::triplifyString($id, parent::getVoc() . "modifier", $modifier) . parent::triplify($uri, parent::getVoc() . strtolower($k), $id));
preg_match_all("/ \\[([^\\]]+)\\]/", $v, $m);
if (isset($m[1])) {
foreach ($m[1] as $item) {
if (!strstr($item, "KO")) {
$item = "kegg:" . str_replace(":", "_", $item);
} else {
$item = str_replace("KO:", "kegg:", $item);
}
parent::addRDF(parent::triplify($id, parent::getVoc() . "link", $item));
}
}
continue;
}
// skip these
if (in_array($k, array("ATOM", "BOND", "BRITE", "AASEQ", "NTSEQ", "SEQUENCE"))) {
continue;
}
// simple strings to keep as is
if (in_array($k, array("EXACT_MASS", "FORMULA", "MOL_WEIGHT", "LINEAGE", "LENGTH", "MASS", "COMPOSITION", "NODE", "EDGE", "POSITION"))) {
parent::addRDF(parent::triplifyString($uri, parent::getVoc() . strtolower($k), $v));
continue;
}
// default catchall
parent::addRDF(parent::triplifyString($uri, parent::getVoc() . strtolower($k), $v . " [script:default]"));
}
if (isset($e['reference'])) {
foreach ($e['reference'] as $i => $r) {
$ref = parent::getRes() . $e['id'] . ".ref.{$i}";
parent::addRDF(parent::describeIndividual($ref, $r['title'], parent::getVoc() . "Reference") . parent::describeClass(parent::getVoc() . "Reference", "Reference") . parent::triplifyString($ref, parent::getVoc() . "authors", $r['authors']) . parent::triplifyString($ref, parent::getVoc() . "journal", $r['journal']) . parent::triplify($uri, parent::getVoc() . "reference", $ref));
if (isset($r['pubmed'])) {
parent::addRDF(parent::triplify($ref, parent::getVoc() . "x-pubmed", $r['pubmed']));
}
}
}
fclose($fp);
}
function ParseEntry($obj, $type)
{
$o = $obj["omim"]["entryList"][0]["entry"];
$omim_id = $o['mimNumber'];
$omim_uri = parent::getNamespace() . $o['mimNumber'];
if (isset($o['version'])) {
parent::setDatasetVersion($o['version']);
}
// add the links
parent::addRDF($this->QQuadO_URL($omim_uri, "rdfs:seeAlso", "http://omim.org/entry/" . $omim_id));
parent::addRDF($this->QQuadO_URL($omim_uri, "owl:sameAs", "http://identifiers.org/omim/" . $omim_id));
// parse titles
$titles = $o['titles'];
parent::addRDF(parent::describeIndividual($omim_uri, $titles['preferredTitle'], parent::getVoc() . str_replace(array(" ", "/"), "-", ucfirst($type))) . parent::describeClass(parent::getVoc() . str_replace(array(" ", "/"), "-", ucfirst($type)), $type));
if (isset($titles['preferredTitle'])) {
parent::addRDF(parent::triplifyString($omim_uri, parent::getVoc() . "preferred-title", $titles['preferredTitle']));
}
if (isset($titles['alternativeTitles'])) {
$b = explode(";;", $titles['alternativeTitles']);
foreach ($b as $title) {
parent::addRDF(parent::triplifyString($omim_uri, parent::getVoc() . "alternative-title", trim($title)));
}
}
// parse text sections
if (isset($o['textSectionList'])) {
foreach ($o['textSectionList'] as $i => $section) {
if ($section['textSection']['textSectionTitle'] == "Description") {
parent::addRDF(parent::triplifyString($omim_uri, "dc:description", $section['textSection']['textSectionContent']));
} else {
$p = str_replace(" ", "-", strtolower($section['textSection']['textSectionTitle']));
parent::addRDF(parent::triplifyString($omim_uri, parent::getVoc() . "{$p}", $section['textSection']['textSectionContent']));
}
// parse the omim references
preg_match_all("/\\{([0-9]{6})\\}/", $section['textSection']['textSectionContent'], $m);
if (isset($m[1][0])) {
foreach ($m[1] as $oid) {
parent::addRDF(parent::triplify($omim_uri, parent::getVoc() . "refers-to", "omim:{$oid}"));
}
}
}
}
// allelic variants
if (isset($o['allelicVariantList'])) {
foreach ($o['allelicVariantList'] as $i => $v) {
$v = $v['allelicVariant'];
$uri = parent::getRes() . "{$omim_id}" . "_allele_" . $i;
$label = str_replace("\n", " ", $v['name']);
parent::addRDF(parent::describeIndividual($uri, $label, parent::getVoc() . "Allelic-Variant") . parent::describeClass(parent::getVoc() . "Allelic-Variant", "Allelic Variant"));
if (isset($v['alternativeNames'])) {
$names = explode(";;", $v['alternativeNames']);
foreach ($names as $name) {
$name = str_replace("\n", " ", $name);
parent::addRDF(parent::triplifyString($uri, parent::getVoc() . "alternative-names", $name));
}
}
if (isset($v['text'])) {
parent::addRDF(parent::triplifyString($uri, "dc:description", $v['text']));
}
if (isset($v['mutations'])) {
parent::addRDF(parent::triplifyString($uri, parent::getVoc() . "mutation", $v['mutations']));
}
if (isset($v['dbSnps'])) {
$snps = explode(",", $v['dbSnps']);
foreach ($snps as $snp) {
parent::addRDF(parent::triplify($uri, parent::getVoc() . "x-dbsnp", "dbsnp:" . $snp));
}
}
parent::addRDF(parent::triplify($omim_uri, parent::getVoc() . "variant", $uri));
}
}
// clinical synopsis
if (isset($o['clinicalSynopsis'])) {
$cs = $o['clinicalSynopsis'];
$cs_uri = parent::getRes() . "" . $omim_id . "_cs";
parent::addRDF(parent::describeIndividual($cs_uri, "Clinical synopsis for omim {$omim_id}", parent::getVoc() . "Clinical-Synopsis") . parent::describeClass(parent::getVoc() . "Clinical-Synopsis", "Clinical Synopsis") . parent::triplify($omim_uri, parent::getVoc() . "clinical-synopsis", $cs_uri));
foreach ($cs as $k => $v) {
if (!strstr($k, "Exists")) {
// ignore the boolean assertion.
// @todo ignore provenance for now
if (in_array($k, array('contributors', 'creationDate', 'editHistory', 'epochCreated', 'dateCreated', 'epochUpdated', 'dateUpdated'))) {
continue;
}
if (!is_array($v)) {
$v = array($k => $v);
}
foreach ($v as $k1 => $v1) {
$phenotypes = explode(";", $v1);
foreach ($phenotypes as $coded_phenotype) {
// parse out the codes
$coded_phenotype = trim($coded_phenotype);
if (!$coded_phenotype) {
continue;
}
$phenotype = preg_replace("/\\{.*\\}/", "", $coded_phenotype);
$phenotype_id = parent::getRes() . "" . md5(strtolower($phenotype));
$entity_id = parent::getRes() . "" . $k1;
parent::addRDF(parent::describeIndividual($phenotype_id, $phenotype, parent::getVoc() . 'Characteristic') . parent::describeClass(parent::getVoc() . 'Characteristic', 'Characteristic') . parent::triplify($cs_uri, parent::getVoc() . "feature", $phenotype_id) . parent::describeIndividual($entity_id, $k1, parent::getVoc() . "Entity") . parent::describeClass(parent::getVoc() . "Entity", "Entity") . parent::triplify($phenotype_id, parent::getVoc() . "characteristic-of", $entity_id));
// parse out the vocab references
preg_match_all("/\\{([0-9A-Za-z \\:\\-\\.]+)\\}|;/", $coded_phenotype, $codes);
//preg_match_all("/((UMLS|HPO HP|SNOMEDCT|ICD10CM|ICD9CM|EOM ID)\:[A-Z0-9]+)/",$coded_phenotype,$m);
if (isset($codes[1][0])) {
foreach ($codes[1] as $entry) {
$entries = explode(" ", trim($entry));
foreach ($entries as $e) {
if ($e == "HPO" || $e == "EOM") {
continue;
}
$this->getRegistry()->parseQName($e, $ns, $id);
if (!isset($ns) || $ns == '') {
$b = explode(".", $id);
$ns = "omim";
$id = $b[0];
} else {
$ns = str_replace(array("hpo", "id", "icd10cm", "icd9cm", "snomedct"), array("hp", "eom", "icd10", "icd9", "snomed"), $ns);
}
parent::addRDF(parent::triplify($phenotype_id, parent::getVoc() . "x-{$ns}", "{$ns}:{$id}"));
}
// foreach
}
// foreach
}
// codes
}
//foreach
}
// foreach
}
// exists
}
}
// clinical synopsis
// genemap
if (isset($o['geneMap'])) {
$map = $o['geneMap'];
if (isset($map['chromosome'])) {
parent::addRDF(parent::triplifyString($omim_uri, parent::getVoc() . "chromosome", (string) $map['chromosome']));
}
if (isset($map['cytoLocation'])) {
parent::addRDF(parent::triplifyString($omim_uri, parent::getVoc() . "cytolocation", (string) $map['cytoLocation']));
}
if (isset($map['geneSymbols'])) {
$b = preg_split("/[,;\\. ]+/", $map['geneSymbols']);
foreach ($b as $symbol) {
parent::addRDF(parent::triplify($omim_uri, parent::getVoc() . "gene-symbol", "symbol:" . trim($symbol)));
}
}
if (isset($map['geneName'])) {
$b = explode(",", $map['geneName']);
foreach ($b as $name) {
parent::addRDF(parent::triplifyString($omim_uri, parent::getVoc() . "gene-name", trim($name)));
}
}
if (isset($map['mappingMethod'])) {
$b = explode(",", $map['mappingMethod']);
foreach ($b as $c) {
$mapping_method = trim($c);
$method_uri = $this->get_method_type($mapping_method);
if ($method_uri !== false) {
parent::addRDF(parent::triplify($omim_uri, parent::getVoc() . "mapping-method", $method_uri));
}
}
}
if (isset($map['mouseGeneSymbol'])) {
$b = explode(",", $map['mouseGeneSymbol']);
foreach ($b as $c) {
parent::addRDF(parent::triplify($omim_uri, parent::getVoc() . "mouse-gene-symbol", "symbol:" . strtoupper($c)));
}
}
if (isset($map['mouseMgiID'])) {
$b = explode(",", $map['mouseMgiID']);
foreach ($b as $c) {
parent::addRDF(parent::triplify($omim_uri, parent::getVoc() . "x-mgi", $c));
}
}
if (isset($map['geneInheritance']) && $map['geneInheritance'] != '') {
parent::addRDF(parent::triplifyString($omim_uri, parent::getVoc() . "gene-inheritance", $map['geneInheritance']));
}
}
if (isset($o['phenotypeMapList'])) {
foreach ($o['phenotypeMapList'] as $i => $phenotypeMap) {
$phenotypeMap = $phenotypeMap['phenotypeMap'];
$pm_uri = parent::getRes() . $omim_id . "_pm_" . ($i + 1);
parent::addRDF(parent::describeIndividual($pm_uri, "phenotype mapping for {$omim_id}", parent::getVoc() . "Phenotype-Map") . parent::describeClass(parent::getVoc() . "Phenotype-Map", "OMIM Phenotype-Map") . parent::triplify($omim_uri, parent::getVoc() . "phenotype-map", $pm_uri));
foreach (array_keys($phenotypeMap) as $k) {
if (in_array($k, array("mimNumber", "phenotypeMimNumber", "phenotypicSeriesMimNumber"))) {
parent::addRDF(parent::triplify($pm_uri, parent::getVoc() . $k, "omim:" . $phenotypeMap[$k]));
} else {
if ($k == "geneSymbols") {
$l = explode(", ", $phenotypeMap[$k]);
foreach ($l as $gene) {
parent::addRDF(parent::triplify($pm_uri, parent::getVoc() . "gene-symbol", "hgnc.symbol:" . $gene));
}
} else {
if ($k == "phenotypeMappingKey") {
$l = $this->get_phenotype_mapping_method_type($phenotypeMap[$k]);
parent::addRDF(parent::triplify($pm_uri, parent::getVoc() . "mapping-method", $l));
} else {
parent::addRDF(parent::triplifyString($pm_uri, parent::getVoc() . $k, $phenotypeMap[$k]));
}
}
}
}
}
}
// references
if (isset($o['referenceList'])) {
foreach ($o['referenceList'] as $i => $r) {
$r = $r['reference'];
if (isset($r['pubmedID'])) {
$pubmed_uri = "pubmed:" . $r['pubmedID'];
parent::addRDF(parent::triplify($omim_uri, parent::getVoc() . "article", $pubmed_uri));
$title = 'article';
if (isset($r['title'])) {
$title = $r['title'];
}
parent::addRDF(parent::describe($pubmed_uri, addslashes($r['title'])));
if (isset($r['articleUrl'])) {
parent::addRDF($this->QQuadO_URL($pubmed_uri, "rdfs:seeAlso", htmlentities($r['articleUrl'])));
}
}
}
}
// external ids
if (isset($o['externalLinks'])) {
foreach ($o['externalLinks'] as $k => $id) {
if ($id === false) {
continue;
}
$ns = '';
switch ($k) {
case 'approvedGeneSymbols':
$ns = 'symbol';
break;
case 'geneIDs':
$ns = 'ncbigene';
break;
case 'ncbiReferenceSequences':
$ns = 'gi';
break;
case 'genbankNucleotideSequences':
$ns = 'gi';
break;
case 'proteinSequences':
$ns = 'gi';
break;
case 'uniGenes':
$ns = 'unigene';
break;
case 'ensemblIDs':
$ns = 'ensembl';
break;
case 'swissProtIDs':
$ns = 'uniprot';
break;
case 'mgiIDs':
$ns = 'mgi';
$b = explode(":", $id);
$id = $b[1];
break;
case 'flybaseIDs':
$ns = 'flybase';
break;
case 'zfinIDs':
$ns = 'zfin';
break;
case 'hprdIDs':
$ns = 'hprd';
break;
case 'orphanetDiseases':
$ns = 'orphanet';
break;
case 'refSeqAccessionIDs':
$ns = 'refseq';
break;
case 'ordrDiseases':
$ns = 'ordr';
$b = explode(";;", $id);
$id = $b[0];
break;
case 'snomedctIDs':
$ns = 'snomed';
break;
case 'icd10cmIDs':
$ns = 'icd10';
break;
case 'icd9cmIDs':
$ns = 'icd9';
break;
case 'umlsIDs':
$ns = 'umls';
break;
case 'wormbaseIDs':
$ns = 'wormbase';
break;
case 'diseaseOntologyIDs':
$ns = 'do';
break;
// specifically ignorning
// specifically ignorning
case 'geneTests':
case 'cmgGene':
case 'geneticAllianceIDs':
// #
// #
case 'nextGxDx':
case 'nbkIDs':
// NBK1207;;Alport Syndrome and Thin Basement Membrane Nephropathy
// NBK1207;;Alport Syndrome and Thin Basement Membrane Nephropathy
case 'newbornScreeningUrls':
case 'decipherUrls':
case 'geneReviewShortNames':
case 'locusSpecificDBs':
case 'geneticsHomeReferenceIDs':
case 'omiaIDs':
case 'coriellDiseases':
case 'clinicalDiseaseIDs':
case 'possumSyndromes':
case 'keggPathways':
case 'gtr':
case 'gwasCatalog':
case 'mgiHumanDisease':
case 'wormbaseDO':
case 'dermAtlas':
// true/false
break;
default:
echo "unhandled external link {$k} {$id}" . PHP_EOL;
}
$ids = explode(",", $id);
foreach ($ids as $id) {
if ($ns) {
if (strstr($id, ";;") === FALSE) {
parent::addRDF(parent::triplify($omim_uri, parent::getVoc() . "x-{$ns}", $ns . ':' . $id));
} else {
$b = explode(";;", $id);
// multiple ids//names
foreach ($b as $c) {
preg_match("/([a-z])/", $c, $m);
if (!isset($m[1])) {
parent::addRDF(parent::triplify($omim_uri, parent::getVoc() . "x-{$ns}", $ns . ':' . $c));
}
}
}
}
}
}
}
//external links
}
private function names()
{
while ($l = $this->getReadFile()->read(200000)) {
$a = explode("\t|\t", trim($l, "|\t\r\n"));
if (count($a) == 0) {
continue;
}
$taxid = parent::getNamespace() . trim($a[0]);
$name = utf8_encode($a[1]);
$rel = parent::getVoc() . str_replace(" ", "-", $a[3]);
parent::addRDF(parent::triplifyString($taxid, $rel, $name) . parent::triplifyString($taxid, parent::getVoc() . "unique-name", utf8_encode($a[2])));
if ($rel == "scientific-name") {
parent::addRDF(parent::triplifyString($taxid, "dc:title", $name) . parent::triplifyString($taxid, "rdfs:label", $name));
}
$this->writeRDFBufferToWriteFile();
}
//while
}
function genes($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);
}
