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})"));
}
}
}
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
}
