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 ParseStringArray($string)
{
parent::getRegistry()->parseQName($string, $ns, $str);
$rm = $this->ParseIDLabelArray($str);
if ($rm !== null) {
$id = trim($rm["id"]);
$label = trim($rm["label"]);
if ($ns == 'other' || $ns == 'xx') {
$ns = '';
}
if ($ns == 'complex') {
$ns = 'rogid';
}
$returnMe = array();
$returnMe["label"] = $label;
$returnMe["id"] = $id;
$returnMe["ns"] = $ns;
return $returnMe;
} else {
return null;
}
}
function OBO2RDF($abbv)
{
$abbv = strtolower($abbv);
if ($abbv == "doid") {
$abbv = "do";
}
$minimal = parent::getParameterValue('detail') == 'min' ? true : false;
$minimalp = parent::getParameterValue('detail') == 'min+' ? true : false;
$version = parent::getParameterValue("bio2rdf_release");
$tid = '';
$first = true;
$is_a = false;
$is_deprecated = false;
$min = $buf = '';
$ouri = "http://bio2rdf.org/lsr:" . $abbv;
$dataset_uri = $abbv . "_resource:bio2rdf.dataset.{$abbv}.R" . $version;
parent::setGraphURI($dataset_uri);
$buf = parent::triplify($ouri, "rdf:type", "owl:Ontology");
$graph_uri = '<' . parent::getRegistry()->getFQURI(parent::getGraphURI()) . '>';
$bid = 1;
while ($l = parent::getReadFile()->read()) {
$lt = trim($l);
if (strlen($lt) == 0) {
continue;
}
if ($lt[0] == '!') {
continue;
}
if (strstr($l, "[Term]")) {
// first node?
if ($first == true) {
// ignore the first case
$first = false;
} else {
if ($tid != '' && $is_a == false && $is_deprecated == false) {
$t = parent::triplify($tid, "rdfs:subClassOf", "obo_vocabulary:Entity");
$buf .= $t;
$min .= $t;
}
}
$is_a = false;
$is_deprecated = false;
unset($typedef);
$term = '';
$tid = '';
continue;
} else {
if (strstr($l, "[Typedef]")) {
$is_a = false;
$is_deprecated = false;
unset($term);
$tid = '';
$typedef = '';
continue;
}
}
//echo "LINE: $l".PHP_EOL;
// to fix error in obo generator
$lt = str_replace("synonym ", "synonym: ", $lt);
$lt = preg_replace("/\\{.*\\} !/", " !", $lt);
$a = explode(" !", $lt);
if (isset($a[1])) {
$exc = trim($a[1]);
}
$a = explode(": ", trim($a[0]), 2);
// let's go
if (isset($intersection_of)) {
if ($a[0] != "intersection_of") {
// $intersection_of .= ")].".PHP_EOL;
//$buf .= $intersection_of;
if ($minimalp) {
$min .= $intersection_of;
}
unset($intersection_of);
}
}
if (isset($relationship)) {
if ($a[0] != "relationship") {
// $relationship .= ")].".PHP_EOL;
//$buf .= $relationship;
if ($minimalp) {
$min .= $relationship;
}
unset($relationship);
}
}
if (isset($typedef)) {
if ($a[0] == "id") {
$c = explode(":", $a[1]);
if (count($c) == 1) {
$ns = "obo";
$id = $c[0];
} else {
$ns = strtolower($c[0]);
$id = $c[1];
}
$id = str_replace(array("(", ")"), array("_", ""), $id);
$tid = $ns . ":" . $id;
} else {
if ($a[0] == "name") {
$buf .= parent::describeClass($tid, addslashes(stripslashes($a[1])));
} else {
if ($a[0] == "is_a") {
if (FALSE !== ($pos = strpos($a[1], "!"))) {
$a[1] = substr($a[1], 0, $pos - 1);
}
$buf .= parent::triplify($tid, "rdfs:subPropertyOf", "obo_vocabulary:" . strtolower($a[1]));
} else {
if ($a[0] == "is_obsolete") {
$buf .= parent::triplify($tid, "rdf:type", "owl:DeprecatedClass");
$is_deprecated = true;
} else {
if ($a[0][0] == "!") {
$a[0] = substr($a[0], 1);
}
$buf .= parent::triplifyString($tid, "obo_vocabulary:{$a['0']}", str_replace('"', '', stripslashes($a[1])));
}
}
}
}
} else {
if (isset($term)) {
if ($a[0] == "is_obsolete" && $a[1] == "true") {
$t = parent::triplify($tid, "rdf:type", "owl:DeprecatedClass");
$t .= parent::triplify($tid, "rdfs:subClassOf", "owl:DeprecatedClass");
$min .= $t;
$buf .= $t;
$is_deprecated = true;
} else {
if ($a[0] == "id") {
parent::getRegistry()->parseQName($a[1], $ns, $id);
$tid = "{$ns}:{$id}";
// $buf .= parent::describeClass($tid,null,"owl:Class");
// $buf .= parent::triplify($tid,"rdfs:isDefinedBy",$ouri);
} else {
if ($a[0] == "name") {
// $t = parent::triplifyString($tid,"rdfs:label",str_replace(array("\"", "'"), array("","\\\'"), stripslashes($a[1]))." [$tid]");
$label = str_replace(array("\"", "'"), array("", "\\\\'"), stripslashes($a[1]));
$t = parent::describeIndividual($tid, $label, "owl:Class");
$t .= parent::triplify($tid, "rdfs:isDefinedBy", $ouri);
$min .= $t;
$buf .= $t;
} else {
if ($a[0] == "def") {
$t = str_replace(array("'", "\"", "\\", "\\\\'"), array("\\\\'", "", "", ""), $a[1]);
$min .= parent::triplifyString($tid, "dc:description", $t);
$buf .= parent::triplifyString($tid, "dc:description", $t);
} else {
if ($a[0] == "property_value") {
$b = explode(" ", $a[1]);
$buf .= parent::triplifyString($tid, "obo_vocabulary:" . strtolower($b[0]), str_replace("\"", "", strtolower($b[1])));
} else {
if ($a[0] == "xref") {
// http://upload.wikimedia.org/wikipedia/commons/3/34/Anatomical_Directions_and_Axes.JPG
// Medical Dictionary:http\://www.medterms.com/
// KEGG COMPOUND:C02788 "KEGG COMPOUND"
// id-validation-regexp:\"REACT_[0-9\]\{1\,4}\\.[0-9\]\{1\,3}|[0-9\]+\"
//$a[1] = 'id-validation-regexp:\"REACT_[0-9\]\{1\,4}\\.[0-9\]\{1\,3}|[0-9\]+\"';
if (substr($a[1], 0, 4) == "http") {
$buf .= parent::triplify($tid, "rdfs:seeAlso", str_replace(array(" ", '"wiki"', "\\"), array("+", "", ""), $a[1]));
} else {
$b = explode(":", $a[1], 2);
if (substr($b[1], 0, 4) == "http") {
$buf .= parent::triplify($tid, "rdfs:seeAlso", stripslashes($b[1]));
} else {
$ns = str_replace(array(" ", "\\"), "", strtolower($b[0]));
$id = trim($b[1]);
// there may be a comment to remove
if (FALSE !== ($pos = strrpos($id, ' "'))) {
$comment = substr($id, $pos + 1, -1);
$id = substr($id, 0, $pos);
}
$id = stripslashes($id);
// there may be a source statement to remove
$id = preg_replace("/{.*\\}/", "", $id);
if ($ns == "pmid") {
$ns = "pubmed";
$y = explode(" ", $id);
$id = $y[0];
}
if ($ns == "xx") {
continue;
}
if ($ns == "icd9cm") {
$y = explode(" ", $id);
$id = $y[0];
}
if ($ns == "xref; umls_cui") {
continue;
}
if ($ns == "submitter") {
$ns = "chebi.submitter";
}
if ($ns == "wikipedia" || $ns == "mesh") {
$id = str_replace(" ", "+", $id);
}
if ($ns == "id-validation-regexp") {
$buf .= parent::triplifyString($tid, "obo_vocabulary:{$ns}", addslashes($id));
} else {
$buf .= parent::triplify($tid, "obo_vocabulary:x-{$ns}", "{$ns}:" . str_replace(" ", "-", $id));
}
}
}
} else {
if ($a[0] == "synonym") {
// synonym: "entidades moleculares" RELATED [IUPAC:]
// synonym: "molecular entity" EXACT IUPAC_NAME [IUPAC:]
// synonym: "Chondrococcus macrosporus" RELATED synonym [NCBITaxonRef:Krzemieniewska_and_Krzemieniewski_1926]
//grab string inside double quotes
preg_match('/"(.*)"(.*)/', $a[1], $matches);
if (!empty($matches)) {
$a[1] = str_replace(array("\\", "\"", "'"), array("", "", "\\\\'"), $matches[1] . $matches[2]);
} else {
$a[1] = str_replace(array("\"", "'"), array("", "\\\\'"), $a[1]);
}
$rel = "SYNONYM";
$list = array("EXACT", "BROAD", "RELATED", "NARROW");
$found = false;
foreach ($list as $keyword) {
// get everything after the keyword up until the bracket [
if (FALSE !== ($k_pos = strpos($a[1], $keyword))) {
$str_len = strlen($a[1]);
$keyword_len = strlen($keyword);
$keyword_end_pos = $k_pos + $keyword_len;
$b1_pos = strrpos($a[1], "[");
$b2_pos = strrpos($a[1], "]");
$b_text = substr($a[1], $b1_pos + 1, $b2_pos - $b1_pos - 1);
$diff = $b1_pos - $keyword_end_pos - 1;
if ($diff != 0) {
// then there is more stuff here
$k = substr($a[1], $keyword_end_pos + 1, $diff);
$rel = trim($k);
} else {
// create the long predicate
$rel = $keyword . "_SYNONYM";
}
$found = true;
$str = substr($a[1], 0, $k_pos - 1);
break;
}
}
// check to see if we still haven't found anything
if ($found === false) {
// we didn't find one of the keywords
// so take from the start to the bracket
$b1_pos = strrpos($a[1], "[");
$str = substr($a[1], 0, $b1_pos - 1);
}
$rel = str_replace(" ", "_", $rel);
// $lit = addslashes($str.($b_text?" [".$b_text."]":""));
$l = parent::triplifyString($tid, "obo_vocabulary:" . strtolower($rel), $str);
$buf .= $l;
} else {
if ($a[0] == "alt_id") {
parent::getRegistry()->parseQname($a[1], $ns, $id);
if ($id != 'curators') {
$buf .= parent::triplify("{$ns}:{$id}", "rdfs:seeAlso", stripslashes($tid));
}
} else {
if ($a[0] == "is_a") {
// do subclassing
parent::getRegistry()->parseQName($a[1], $ns, $id);
$t = parent::triplify($tid, "rdfs:subClassOf", "{$ns}:{$id}");
$buf .= $t;
$min .= $t;
$is_a = true;
} else {
if ($a[0] == "intersection_of") {
if (!isset($intersection_of)) {
// $intersection_of = '<'.parent::getRegistry()->getFQURI($tid).'> <'.parent::getRegistry()->getFQURI('owl:equivalentClass').'> [<'.parent::getRegistry()->getFQURI('rdf:type').'> <'.parent::getRegistry()->getFQURI('owl:Class').'>; <'.parent::getRegistry()->getFQURI('owl:intersectionOf').'> (';
$intersection_of = '<' . parent::getRegistry()->getFQURI($tid) . '> <' . parent::getRegistry()->getFQURI('owl:equivalentClass') . '> _:b' . ++$bid . " {$graph_uri} ." . PHP_EOL;
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('rdf:type') . '> <' . parent::getRegistry()->getFQURI('owl:Class') . "> {$graph_uri} ." . PHP_EOL;
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:intersectionOf') . '> _:b' . ++$bid . " {$graph_uri} ." . PHP_EOL;
}
/*
intersection_of: ECO:0000206 ! BLAST evidence
intersection_of: develops_from VAO:0000092 ! chondrogenic condensation
intersection_of: OBO_REL:has_part VAO:0000040 ! cartilage tissue
*/
$c = explode(" ", $a[1]);
if (count($c) == 1) {
// just a class
parent::getRegistry()->parseQName($c[0], $ns, $id);
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('rdfs:subClassOf') . '> <' . parent::getRegistry()->getFQURI("{$ns}:{$id}") . "> {$graph_uri} ." . PHP_EOL;
$buf .= parent::triplify($tid, "rdfs:subClassOf", "{$ns}:{$id}");
} else {
if (count($c) == 2) {
// an expression
parent::getRegistry()->parseQName($c[0], $pred_ns, $pred_id);
parent::getRegistry()->parseQName($c[1], $obj_ns, $obj_id);
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:onProperty') . '> <' . parent::getRegistry()->getFQURI("obo_vocabulary:" . $pred_id) . "> {$graph_uri} ." . PHP_EOL;
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:someValuesFrom') . '> <' . parent::getRegistry()->getFQURI("{$obj_ns}:{$obj_id}") . "> {$graph_uri} ." . PHP_EOL;
$buf .= parent::triplify($tid, "obo_vocabulary:{$pred_id}", "{$obj_ns}:{$obj_id}");
}
}
} else {
if ($a[0] == "relationship") {
if (!isset($relationship)) {
$relationship = '<' . parent::getRegistry()->getFQURI($tid) . '> <' . parent::getRegistry()->getFQURI('rdfs:subClassOf') . '> _:b' . ++$bid . " {$graph_uri} ." . PHP_EOL;
$relationship .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('rdf:type') . '> <' . parent::getRegistry()->getFQURI('owl:Class') . "> {$graph_uri} ." . PHP_EOL;
$relationship .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:intersectionOf') . '> _:b' . ++$bid . " {$graph_uri} ." . PHP_EOL;
}
/*
relationship: develops_from VAO:0000092 ! chondrogenic condensation
relationship: OBO_REL:has_part VAO:0000040 ! cartilage tissue
*/
$c = explode(" ", $a[1]);
if (count($c) == 1) {
// just a class
parent::getRegistry()->parseQName($c[0], $ns, $id);
$relationship .= parent::getRegistry()->getFQURI("{$ns}:{$id}");
$buf .= parent::triplify($tid, "rdfs:subClassOf", "{$ns}:{$id}");
} else {
if (count($c) == 2) {
// an expression
parent::getRegistry()->parseQName($c[0], $pred_ns, $pred_id);
parent::getRegistry()->parseQName($c[1], $obj_ns, $obj_id);
$relationship .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:onProperty') . '> <' . parent::getRegistry()->getFQURI("obo_vocabulary:" . $pred_id) . "> {$graph_uri} ." . PHP_EOL;
$relationship .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:someValuesFrom') . '> <' . parent::getRegistry()->getFQURI("{$obj_ns}:{$obj_id}") . "> {$graph_uri} ." . PHP_EOL;
$buf .= parent::triplify($tid, "obo_vocabulary:{$pred_id}", "{$obj_ns}:{$obj_id}");
}
}
} else {
// default handler
if (isset($a[1])) {
$buf .= parent::triplifyString($tid, "obo_vocabulary:{$a['0']}", str_replace(array("\"", "'"), array("", "\\\\'"), stripslashes($a[1])));
}
}
}
}
}
}
}
}
}
}
}
}
} else {
//header
//format-version: 1.0
$buf .= parent::triplifyString($ouri, "obo_vocabulary:{$a['0']}", str_replace(array('"', '\\:'), array('\\"', ':'), isset($a[1]) ? $a[1] : ""));
}
}
if ($minimal || $minimalp) {
parent::getWriteFile()->write($min);
} else {
parent::getWriteFile()->write($buf);
}
$min = '';
$buf = '';
$header = '';
}
//if(isset($intersection_of)) $buf .= $intersection_of.")].".PHP_EOL;
//if(isset($relationship)) $buf .= $relationship.")].".PHP_EOL;
if ($minimal || $minimalp) {
parent::getWriteFile()->Write($min);
} else {
parent::getWriteFile()->write($buf);
}
}