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);
}
function functional_descriptions()
{
while ($l = $this->getReadFile()->read(2000000)) {
if ($l[0] == "#") {
continue;
}
if (strstr($l, "gene_id")) {
continue;
}
// gene_id public_name molecular_name concise_description provisional_description detailed_description automated_description gene_class_description
$a = explode("\t", $l);
if (count($a) != 8) {
trigger_error("Found one row that only has " . count($a) . " columns, expecting 8", E_USER_ERROR);
continue;
}
$id = parent::getNamespace() . $a[0];
$label = $a[1] . ($a[2] ? " (" . $a[2] . ")" : "");
parent::addRDF(parent::describeIndividual($id, $label, parent::getVoc() . "Gene") . parent::describeClass(parent::getVoc() . "Gene", "Wormbase Gene") . parent::triplifyString($id, parent::getVoc() . "concise-description", $a[3]) . parent::triplifyString($id, parent::getVoc() . "provisional-description", $a[4]) . parent::triplifyString($id, parent::getVoc() . "detailed-description", $a[5]) . parent::triplifyString($id, parent::getVoc() . "automated-description", $a[6]) . parent::triplifyString($id, parent::getVoc() . "gene-class-description", trim($a[7])));
parent::writeRDFBufferToWriteFile();
}
}
public function makeAddress($address)
{
if ($address == null) {
return null;
}
$address_uri = parent::getRes() . md5($address->asXML());
parent::addRDF(parent::describeIndividual($address_uri, "address", parent::getVoc() . "Address") . parent::describeClass(parent::getVoc() . "Address", "Address") . parent::triplifyString($address_uri, parent::getVoc() . "city", $this->makeDescription($this->getString('./city', $address), "City")) . parent::triplifyString($address_uri, parent::getVoc() . "state", $this->makeDescription($this->getString('./state', $address), "State")) . parent::triplifyString($address_uri, parent::getVoc() . "zip", $this->makeDescription($this->getString('./zip', $address), "ZipCode")) . parent::triplifyString($address_uri, parent::getVoc() . "country", $this->makeDescription($this->getString('./country', $address), "Country")));
return $address_uri;
}
function product($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 process()
{
$refseq_record_str = "";
while ($aLine = $this->getReadFile()->Read(40960)) {
preg_match("/^\\/\\/\$/", $aLine, $matches);
if (!count($matches)) {
preg_match("/^\n\$/", $aLine, $matches);
if (count($matches) == 0) {
$refseq_record_str .= $aLine . PHP_EOL;
}
continue;
} else {
//now remove the header if it is there
$refseq_record_str = $this->removeHeader($refseq_record_str);
$sectionsRaw = $this->parseGenbankRaw($refseq_record_str);
/**
* SECTIONS being parsed:
* locus, definition, accession, version, keywords, source
* features
**/
//get the locus section
$locus = $this->retrieveSections("LOCUS", $sectionsRaw);
$parsed_locus_arr = $this->parseLocus($locus);
//get the definition
$definition = $this->retrieveSections("DEFINITION", $sectionsRaw);
$parsed_definition_arr = $this->parseDefinition($definition);
//get the accession
$accessions = $this->retrieveSections("ACCESSION", $sectionsRaw);
$parsed_accession_arr = $this->parseAccession($accessions);
//get the version
$versions = $this->retrieveSections("VERSION", $sectionsRaw);
$parsed_version_arr = $this->parseVersion($versions);
//get the keywords
$keywords = $this->retrieveSections("KEYWORDS", $sectionsRaw);
$parsed_keyword_arr = $this->parseKeywords($keywords);
//get the reference section
$references = $this->retrieveSections("REFERENCE", $sectionsRaw);
$parsed_refs_arr = $this->parseReferences($references);
//get the source section
$source = $this->retrieveSections("SOURCE", $sectionsRaw);
$parsed_source_arr = $this->parseSource($source);
//get the features
$features = $this->retrieveSections("FEATURES", $sectionsRaw);
$parsed_features_arr = $this->parseFeatures($features);
//lets make some rdf
$refseq_res = $this->getNamespace() . $parsed_version_arr['versioned_accession'];
$refseq_label = utf8_encode(htmlspecialchars($parsed_definition_arr[0]));
parent::AddRDF(parent::describeIndividual($refseq_res, $refseq_label, $this->getVoc() . 'refseq-record') . parent::triplifyString($refseq_res, $this->getVoc() . 'sequence-length', $parsed_locus_arr[0]['sequence_length']) . parent::triplifyString($refseq_res, $this->getVoc() . 'chromosome-shape', $parsed_locus_arr[0]['chromosome_shape']) . parent::triplifyString($refseq_res, $this->getVoc() . 'date-of-entry', $parsed_locus_arr[0]['date']) . parent::triplifyString($refseq_res, $this->getVoc() . 'source', utf8_encode($parsed_source_arr[0])) . parent::triplify($refseq_res, $this->getVoc() . 'fasta-seq', 'https://www.ncbi.nlm.nih.gov/sviewer/viewer.cgi?sendto=on&db=nucest&dopt=fasta&val=' . $parsed_version_arr['gi']) . parent::triplify('https://www.ncbi.nlm.nih.gov/sviewer/viewer.cgi?sendto=on&db=nucest&dopt=fasta&val=' . $parsed_version_arr['gi'], "rdf:type", $this->getVoc() . 'fasta-sequence'));
//add the features to the rdf
foreach ($parsed_features_arr as $aFeature) {
$type = $aFeature['type'];
$feat_desc = $this->getFeatures($type);
$label = $type;
$def = '';
if (isset($feat_desc['definition'])) {
$def = preg_replace('/\\s\\s*/', ' ', $feat_desc['definition']);
}
$comment = null;
// $value = $aFeature['value'];
$value = str_replace("UniProtKB/Swiss-Prot", "UniProt", $aFeature['value']);
// imperfect solution.
$value_arr = explode("/", $value);
$location = preg_replace('/\\n/', '', $value_arr[0]);
$class_id = parent::getVoc() . md5($type);
$feat_res = parent::getRes() . md5($type . $location . $refseq_res);
$feat_label = utf8_encode($type . " " . $location . " for " . $refseq_res);
if (isset($feat_desc['comment'])) {
$comment = $feat_desc['comment'];
$comment = preg_replace('/\\s\\s*/', ' ', $comment);
$label .= " " . $comment;
}
parent::AddRDF(parent::describeClass($class_id, $label, parent::getVoc() . "Feature", $label, $def) . parent::describeIndividual($feat_res, $feat_label, $class_id) . parent::triplify($refseq_res, $this->getVoc() . "has-feature", $feat_res));
foreach ($value_arr as $aL) {
//check if aL has an equals in it
$p = "/(\\S+)\\=(.*)/";
preg_match($p, $aL, $m);
if (count($m)) {
if ($m[1] == "db_xref") {
parent::AddRDF(parent::triplify($feat_res, "rdfs:seeAlso", str_replace("\"", "", $m[2])));
} else {
parent::AddRDF(parent::triplifyString($feat_res, $this->getVoc() . $m[1], utf8_encode(str_replace("\"", "", $m[2]))));
}
}
}
}
//add the accession
foreach ($parsed_accession_arr[0] as $acc) {
parent::AddRDF(parent::triplifyString($refseq_res, $this->getVoc() . "accession", $acc));
}
//versioned accession
if (isset($parsed_version_arr['versioned_accession'])) {
parent::AddRDF(parent::triplifyString($refseq_res, $this->getVoc() . "versioned-accession", $parsed_version_arr['versioned_accession']));
}
//keywords
foreach ($parsed_keyword_arr as $akw) {
parent::AddRDF(parent::triplifyString($refseq_res, $this->getVoc() . "keyword", $akw));
}
//references
foreach ($parsed_refs_arr as $aRef) {
$r = rand();
$ref_res = $this->getRes() . md5($r);
$ref_label = "reference for " . $refseq_res;
if (isset($aRef['TITLE'])) {
parent::AddRDF(parent::describeIndividual($ref_res, $ref_label, $this->getVoc() . "reference") . parent::triplifyString($ref_res, $this->getVoc() . "title", $aRef['TITLE']));
}
if (isset($aRef['PUBMED'])) {
parent::AddRDF(parent::triplify($ref_res, $this->getVoc() . "x-pubmed", 'pubmed:' . $aRef['PUBMED']));
}
if (isset($aRef['AUTHORS'])) {
parent::AddRDF(parent::triplifyString($ref_res, $this->getVoc() . "authors", $aRef['AUTHORS']));
}
if (isset($aRef['COORDINATES'])) {
parent::AddRDF(parent::triplify($refseq_res, $this->getVoc() . "reference", $ref_res) . parent::triplifyString($ref_res, $this->getVoc() . "coordinates", $aRef['COORDINATES']) . parent::triplifyString($ref_res, $this->getVoc() . "citation", $aRef['JOURNAL']));
} else {
parent::AddRDF(parent::triplify($refseq_res, $this->getVoc() . "reference", $ref_res) . parent::triplifyString($ref_res, $this->getVoc() . "citation", $aRef['JOURNAL']));
}
}
$refseq_record_str = "";
$this->WriteRDFBufferToWriteFile();
continue;
}
}
//while
}
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 twosides()
{
$items = null;
$id = 0;
$this->GetReadFile()->Read();
while ($l = $this->GetReadFile()->Read()) {
$a = explode("\t", $l);
$id++;
if ($id % 10000 == 0) {
$this->WriteRDFBufferToWriteFile();
}
$uid = "twosides:{$id}";
$d1 = "pubchemcompound:" . (int) sprintf("%d", substr($a[0], 4));
$d1_name = $a[2];
$d2 = "pubchemcompound:" . (int) sprintf("%d", substr($a[1], 4));
$d2_name = $a[3];
$e = "umls:" . $a[4];
$e_name = strtolower($a[5]);
$uid_label = "DDI between {$d1_name} and {$d2_name} leading to {$e_name}";
if (!isset($items[$d1])) {
parent::addRDF(parent::describeIndividual($d1, $d1_name, parent::getVoc() . "Chemical") . parent::describeClass(parent::getVoc() . "Chemical", "PharmGKB Chemical"));
$items[$d1] = '';
}
if (!isset($items[$d2])) {
parent::addRDF(parent::describeIndividual($d2, $d2_name, parent::getVoc() . "Chemical") . parent::describeClass(parent::getVoc() . "Chemical", "PharmGKB Chemical"));
$items[$d2] = '';
}
if (!isset($items[$e])) {
parent::addRDF(parent::describeIndividual($e, $e_name, parent::getVoc() . "Event") . parent::describeClass(parent::getVoc() . "Event", "PharmGKB side effect event"));
$items[$e] = '';
}
parent::addRDF(parent::describeIndividual($uid, $uid_label, parent::getVoc() . "Drug-Drug-Association") . parent::describeClass(parent::getVoc() . "Drug-Drug-Association", "PharmGKB Twosides Drug-Drug Association") . parent::triplify($uid, parent::getVoc() . "chemical", $d1) . parent::triplify($uid, parent::getVoc() . "chemical", $d2) . parent::triplify($uid, parent::getVoc() . "event", $e) . parent::triplifyString($uid, parent::getVoc() . "p-value", $a[7]));
}
parent::writeRDFBufferToWriteFile();
}
function ParseEntry($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
}
function Parse()
{
$l = parent::getReadFile()->read(100000);
$header = explode("\t", trim(substr($l, 1)));
if (($c = count($header)) != 54) {
trigger_erorr("Expecting 54 columns, found {$c}!");
return FALSE;
}
// check # of columns
while ($l = parent::getReadFile()->read(500000)) {
$a = explode("\t", trim($l));
// irefindex identifiers
$rigid = "irefindex." . $a[34];
# checksum for interaction
$rogida = "irefindex." . $a[32];
# checksum for A
$rogidb = "irefindex." . $a[33];
# checksum for B
$irigid = "irefindex.irigid:" . $a[44];
# integer id for interaction
$irogida = "irefindex.irogid:" . $a[42];
# integer id for A
$irogidb = "irefindex.irogid:" . $a[43];
# integer id for B
$crigid = "irefindex.crigid:" . $a[47];
# checksum for canonical interaction
$icrigid = "irefindex.icrigid:" . $a[50];
# integer id for canonical interaction
$crogida = "irefindex.crogid:" . $a[45];
# checksum for A's canonical group
$crogidb = "irefindex.crogid:" . $a[46];
# checksum for B's canonical group
$icrogida = "irefindex.icrogid:" . $a[48];
# integer for A's canonical group
$icrogidb = "irefindex.icrogid:" . $a[49];
# integer for B's canonical group
// 13 contains the original identifier, the rigid, and the edgetype
$ids = explode("|", $a[13]);
if (count($ids) != 3) {
trigger_error("Expecting 3 entries in column 14");
print_r($ids);
exit;
}
parent::getRegistry()->parseQName($ids[0], $ns, $id);
if ($id == '-') {
// this happens with hprd
$iid = "hprd:" . substr($ids[1], 6);
} else {
$iid = $ns . ":" . $id;
}
// get the type
if ($a[52] == "X") {
$label = "{$a['0']} - {$a['1']} Interaction";
$type = "Pairwise-Interaction";
} else {
if ($a[52] == "C") {
$label = $a[53] . " component complex";
#num of participants
$type = "Multimeric-Complex";
} else {
if ($a[52] == "Y") {
$label = "{$a['0']} homomeric complex";
$type = "Homopolymeric-Complex";
}
}
}
parent::addRDF(parent::describeIndividual($iid, $label, parent::getVoc() . $type) . parent::describeClass(parent::getVoc() . $type, str_replace("-", " ", $type)));
// interaction type[52] by method[6]
unset($method);
if ($a[6] != '-') {
$data = $this->ParseStringArray($a[6]);
$method = trim($data["label"]);
$qname = trim($data["ns"]) . ":" . trim($data["id"]);
if ($qname) {
parent::addRDF(parent::triplify($iid, parent::getVoc() . "method", $qname) . parent::describeClass($qname, $data['label']));
}
}
parent::addRDF(parent::triplify($iid, "rdfs:seeAlso", "http://wodaklab.org/iRefWeb/interaction/show/" . $a[50]));
// set the interactors
for ($i = 0; $i <= 1; $i++) {
$p = 'a';
if ($i == 1) {
$p = 'b';
}
$data = $this->ParseStringArray($a[$i]);
$interactor = trim($data["ns"]) . ":" . trim($data["id"]);
parent::addRDF(parent::triplify($iid, parent::getVoc() . "interactor_{$p}", $interactor));
// biological role
$role = $a[16 + $i];
if ($role != '-') {
$data = $this->ParseStringArray($role);
$qname = trim($data["ns"]) . ":" . trim($data["id"]);
if ($qname != "mi:0000") {
parent::addRDF(parent::triplify($iid, parent::getVoc() . "interactor_{$p}" . "_biological_role", $qname) . parent::describeClass($qname, $data['label']));
}
}
// experimental role
$role = $a[18 + $i];
if ($role != '-') {
$data = $this->ParseStringArray($role);
$qname = trim($data["ns"]) . ":" . trim($data["id"]);
if ($qname != "mi:0000") {
parent::addRDF(parent::triplify($iid, parent::getVoc() . "interactor_{$p}" . "_experimental_role", $qname) . parent::describeClass($qname, $data['label']));
}
}
// interactor type
$type = $a[20 + $i];
if ($type != '-') {
$data = $this->ParseStringArray($type);
$qname = trim($data["ns"]) . ":" . trim($data["id"]);
parent::addRDF(parent::triplify($interactor, "rdf:type", $qname) . parent::describeClass($qname, $data['label']));
}
}
// add the alternatives through the taxon + seq redundant group
for ($i = 2; $i <= 3; $i++) {
$taxid = '';
$rogid = "irefindex." . $a[32 + ($i - 2)];
parent::addRDF(parent::describeIndividual($rogid, "", parent::getVoc() . "Taxon-Sequence-Identical-Group") . parent::describeClass(parent::getVoc() . "Taxon-Sequence-Identical-Group", "Taxon + Sequence Identical Group"));
$tax = $a[9 + ($i - 2)];
if ($tax && $tax != '-' && $tax != '-1') {
$data = $this->ParseStringArray($tax);
$taxid = trim($data["ns"]) . ":" . trim($data["id"]);
parent::addRDF(parent::triplify($rogid, parent::getVoc() . "x-taxonomy", $taxid));
}
$list = explode("|", $a[3 + ($i - 2)]);
foreach ($list as $item) {
$data = $this->ParseStringArray($item);
$ns = trim($data["ns"]);
$id = trim($data["id"]);
$qname = $ns . ":" . $id;
if ($ns && $ns != 'rogid' && $ns != 'irogid' and $id != '-') {
parent::addRDF(parent::triplify($rogid, parent::getVoc() . "has-member", $qname));
if ($taxid && $taxid != '-' && $taxid != '-1') {
parent::addRDF(parent::triplify($qname, parent::getVoc() . "x-taxonomy", $taxid));
}
}
}
}
// publications
$list = explode("|", $a[8]);
foreach ($list as $item) {
if ($item == '-' && $item != 'pubmed:0') {
continue;
}
$data = $this->ParseStringArray($item);
$qname = trim($data["ns"]) . ":" . trim($data["id"]);
parent::addRDF(parent::triplify($iid, parent::getVoc() . "article", $qname));
}
// MI interaction type
if ($a[11] != '-' && $a[11] != 'NA') {
$data = $this->ParseStringArray($a[11]);
$qname = trim($data["ns"]) . ":" . trim($data["id"]);
parent::addRDF(parent::triplify($iid, "rdf:type", $qname));
if (!isset($defined[$qname])) {
$defined[$qname] = '';
parent::addRDF(parent::triplifyString($qname, "rdfs:label", $data['label']));
}
}
// source
if ($a[12] != '-') {
$data = $this->ParseStringArray($a[12]);
$qname = trim($data["ns"]) . ":" . trim($data["id"]);
parent::addRDF(parent::triplify($iid, parent::getVoc() . "source", $qname));
}
// confidence
$list = explode("|", $a[14]);
foreach ($list as $item) {
$data = $this->ParseStringArray($item);
$ns = trim($data["ns"]);
$id = trim($data["id"]);
if ($ns == 'lpr') {
// lowest number of distinct interactions that any one article reported
parent::addRDF(parent::triplifyString($iid, parent::getVoc() . "minimum-number-interactions-reported", $id));
} else {
if ($ns == "hpr") {
// higher number of distinct interactions that any one article reports
parent::addRDF(parent::triplifyString($iid, parent::getVoc() . "maximum-number-interactions-reported", $id));
} else {
if ($ns = 'hp') {
// total number of unique PMIDs used to support the interaction
parent::addRDF(parent::triplifyString($iid, parent::getVoc() . "number-supporting-articles", $id));
}
}
}
}
// expansion method
if ($a[15]) {
$id = parent::getRes() . md5($a[15]);
parent::addRDF(parent::describeIndividual($id, $a[15], parent::getVoc() . "Expansion-Method") . parent::describeClass(parent::getVoc() . "Expansion-Method", "Expansion Method") . parent::triplify($iid, parent::getVoc() . "expansion-method", $id));
}
// host organism
if ($a[28] != '-') {
$data = $this->ParseStringArray($a[28]);
$qname = trim($data["ns"]) . ":" . trim($data["id"]);
parent::addRDF(parent::triplify($iid, parent::getVoc() . "host-organism", $qname));
}
// @todo add to record
// created 2010/05/18
$date = str_replace("/", "-", $a[30]) . "T00:00:00Z";
parent::addRDF(parent::triplifyString($iid, "dc:created", $date, "xsd:dateTime"));
// taxon-sequence identical interaction group
parent::addRDF(parent::triplify($iid, parent::getVoc() . "taxon-sequence-identical-interaction", $rigid) . parent::triplify($rigid, "rdf:type", parent::getVoc() . "Taxon-Sequence-Identical-Interaction") . parent::describeClass(parent::getVoc() . "Taxon-Sequence-Identical-Interaction", "Taxon + Sequence Identical Interaction") . parent::triplify($rigid, parent::getVoc() . "irigid", $irigid) . parent::triplify($rigid, parent::getVoc() . "interactor-a", $rogida) . parent::triplify($rogida, parent::getVoc() . "irogid", $irogida) . parent::triplify($rigid, parent::getVoc() . "interactor-b", $rogidb) . parent::triplify($rogidb, parent::getVoc() . "irogid", $irogidb) . parent::triplify($rogida, parent::getVoc() . "canonical-group", $crogida) . parent::triplify($rogidb, parent::getVoc() . "canonical-group", $crogidb) . parent::triplify($rigid, parent::getVoc() . "taxon-sequence-similar-interaction", $crigid) . parent::triplify($crigid, "rdf:type", parent::getVoc() . "Taxon-Sequence-Canonical-Interaction") . parent::describeClass(parent::getVoc() . "Taxon-Sequence-Canonical-Interaction", "Taxon + Sequence Canonical Interaction") . parent::triplify($crigid, parent::getVoc() . "icrigid", $icrigid) . parent::triplify($crigid, parent::getVoc() . "interactor-a-canonical-group", $crogida) . parent::triplify($crogida, "rdf:type", parent::getVoc() . "Taxon-Sequence-Similar-Group") . parent::triplify($crogida, parent::getVoc() . "icrogid", $icrogida) . parent::triplify($crigid, parent::getVoc() . "interactor-b-canonical-group", $crogidb) . parent::triplify($crogidb, "rdf:type", parent::getVoc() . "Taxon-Sequence-Similar-Group") . parent::triplify($crogidb, parent::getVoc() . "icrogid", $icrogidb) . parent::describeClass(parent::getVoc() . "Taxon-Sequence-Similar-Group", "Taxon + Sequence Similar Group"));
parent::writeRDFBufferToWriteFile();
}
}
function psiblast()
{
while ($l = $this->GetReadFile()->Read(2048)) {
$a = explode("\t", trim($l));
$id1 = $a[0];
$id2 = $a[7];
$id = "aln_{$id1_}{$id2}";
$this->AddRDF(parent::describeIndividual($this->getRes() . $id, "psiblast alignment between {$id1} and {$id2}", $this->getVoc() . "PSI-BLAST-Alignment") . parent::describeClass($this->getVoc() . "PSI-BLAST-Alignment", "PSI-Blast Alignment") . parent::triplify($this->getRes() . $id, $this->getVoc() . "query", $this->getNamespace() . $id1) . parent::triplify($this->getRes() . $id, $this->getVoc() . "target", $this->getNamespace() . $id2) . parent::triplifyString($this->getRes() . $id, $this->getVoc() . "query-start", $a[1]) . parent::triplifyString($this->getRes() . $id, $this->getVoc() . "query-stop", $a[2]) . parent::triplifyString($this->getRes() . $id, $this->getVoc() . "target-start", $a[3]) . parent::triplifyString($this->getRes() . $id, $this->getVoc() . "target-stop", $a[4]) . parent::triplifyString($this->getRes() . $id, $this->getVoc() . "percent-aligned", $a[5]) . parent::triplifyString($this->getRes() . $id, $this->getVoc() . "score", $a[6]) . parent::triplifyString($this->getRes() . $id, $this->getVoc() . "is-encoded-by", "taxon:" . $a[8]) . parent::describeProperty($this->getVoc() . "target-start", "Relationship between an SGD sequence alignment and its target sequence start position") . parent::describeProperty($this->getVoc() . "target-stop", "Relationship between an SGD sequence alignment and its target sequence stop position") . parent::describeProperty($this->getVoc() . "score", "Relationship between an SGD sequence alignment and its score") . parent::describeProperty($this->getVoc() . "percent-aligned", "Relationship between an SGD sequence alignment and its percent-aligned value") . parent::describeProperty($this->getVoc() . "is-encoded-by", "Relationship between an SGD sequence alignment and the taxon the aligned sequences are encoded by"));
parent::writeRDFBufferToWriteFile();
}
//while
return TRUE;
}
private function citations()
{
while ($l = $this->getReadFile()->read(2000000)) {
$a = explode("\t|\t", rtrim($l, "\t|\n"));
if (!isset($a[1]) or !isset($a[2])) {
continue;
}
$c = parent::getRes() . "citation-id-" . $a[0];
$seealso = isset($a[4]) ? trim($a[4]) : "";
if ($seealso) {
$seealso = str_replace(array("lx: DOI ", "http;//"), array("http://dx.doi.org/", "http://"), $seealso);
if (strlen($seealso) > 2 and !strstr($seealso, "http")) {
$seealso = "http://" . $seealso;
}
$seealso = parent::triplify($c, "rdfs:seeAlso", $seealso);
}
parent::addRDF(parent::describeIndividual($c, $a[1], $this->getVoc() . "Citation") . parent::describeClass($this->getVoc() . "Citation", "Citation") . parent::triplifyString($c, parent::getVoc() . "citation-key", $a[1]) . ($a[2] == "0" ? "" : parent::triplify($c, parent::getVoc() . "x-pubmed", "pubmed:" . $a[2])) . $seealso . ((isset($a[5]) and $a[5]) ? parent::triplifyString($c, parent::getVoc() . "text", str_replace("\"", "", $a[5])) : ""));
if (isset($a[6])) {
$taxids = explode(" ", trim($a[6]));
if (count($taxids)) {
foreach ($taxids as $taxid) {
parent::addRDF(parent::triplify("taxonomy:{$taxid}", $this->getVoc() . "citation", $c));
}
}
}
$this->writeRDFBufferToWriteFile();
}
//while
}
function parseDrugEntry(&$xml)
{
$declared = null;
// a list of all the entities declared
$counter = 1;
$x = $xml->GetXMLRoot();
$dbid = (string) $x->{"drugbank-id"};
$did = "drugbank:" . $dbid;
$name = (string) $x->name;
$type = ucfirst((string) str_replace(" ", "-", $x->attributes()->type));
$type_label = ucfirst($x->attributes()->type);
$description = null;
if (isset($this->id_list)) {
if (!isset($this->id_list[$dbid])) {
return;
}
unset($this->id_list[$dbid]);
}
echo "Processing {$dbid}" . PHP_EOL;
if (isset($x->description) && $x->description != '') {
$description = trim((string) $x->description);
}
parent::addRDF(parent::describeIndividual($did, $name, parent::getVoc() . "Drug", $name, $description) . parent::describeClass(parent::getVoc() . "Drug", "Drug") . parent::triplify($did, "owl:sameAs", "http://identifiers.org/drugbank/" . $dbid) . parent::triplify($did, "rdfs:seeAlso", "http://www.drugbank.ca/drugs/" . $dbid) . parent::triplify($did, "rdf:type", parent::getVoc() . $type) . parent::describeClass(parent::getVoc() . $type, $type_label));
foreach ($x->{'drugbank-id'} as $id) {
parent::addRDF(parent::triplifyString($did, parent::getVoc() . "drugbank-id", $id));
}
if (isset($x->{'cas-number'})) {
parent::addRDF(parent::triplify($did, parent::getVoc() . "x-cas", "cas:" . $x->{'cas-number'}));
}
$literals = array("indication", "pharmacodynamics", "mechanism-of-action", "toxicity", "biotransformation", "absorption", "half-life", "protein-binding", "route-of-elimination", "volume-of-distribution", "clearance");
foreach ($literals as $l) {
if (isset($x->{$l}) and $x->{$l} != '') {
$lid = parent::getRes() . md5($l . $x->{$l});
parent::addRDF(parent::describeIndividual($lid, "{$l} for {$did}", parent::getVoc() . ucfirst($l), "{$l} for {$did}", $x->{$l}) . parent::describeClass(parent::getVoc() . ucfirst($l), ucfirst(str_replace("-", " ", $l))) . parent::triplify($did, parent::getVoc() . $l, $lid));
}
}
// TODO:: Replace the next two lines
$this->AddList($x, $did, "groups", "group", parent::getVoc() . "group");
$this->AddList($x, $did, "categories", "category", parent::getVoc() . "category");
if (isset($x->classification)) {
foreach ($x->classification->children() as $k => $v) {
$cid = parent::getRes() . md5($v);
parent::addRDF(parent::describeIndividual($cid, $v, parent::getVoc() . "Drug-Classification-Category") . parent::describeClass(parent::getVoc() . "Drug-Classification-Category", "Drug Classification Category") . parent::triplify($did, parent::getVoc() . "drug-classification-category", $cid));
}
}
$this->addLinkedResource($x, $did, 'atc-codes', 'atc-code', 'atc');
$this->addLinkedResource($x, $did, 'ahfs-codes', 'ahfs-code', 'ahfs');
// taxonomy
$this->AddText($x, $did, "taxonomy", "kingdom", parent::getVoc() . "kingdom");
// substructures
$this->AddText($x, $did, "taxonomy", "substructures", parent::getVoc() . "substructure", "substructure");
// synonyms
$this->AddCategory($x, $did, "synonyms", "synonym", parent::getVoc() . "synonym");
// brand names
$this->AddCategory($x, $did, "international-brands", "international-brand", parent::getVoc() . "brand");
// salt
if (isset($x->salts->salt)) {
foreach ($x->salts->salt as $s) {
$sid = parent::getPrefix() . ':' . $s->{'drugbank-id'};
parent::addRDF(parent::describeIndividual($sid, $s->name, parent::getVoc() . "Salt") . parent::describeClass(parent::getVoc() . "Salt", "Salt") . parent::triplify($did, parent::getVoc() . "salt", $sid) . parent::triplify($sid, parent::getVoc() . "x-cas", "cas:" . $s->{'cas-number'}) . parent::triplify($sid, parent::getVoc() . "x-inchikey", "inchikey:" . $s->{'inchikey'}));
}
}
// mixtures
// <mixtures><mixture><name>Cauterex</name><ingredients>dornase alfa + fibrinolysin + gentamicin sulfate</ingredients></mixture>
if (isset($x->mixtures)) {
$id = 0;
foreach ($x->mixtures->mixture as $item) {
if (isset($item)) {
$o = $item;
$mid = parent::getRes() . str_replace(" ", "-", $o->name[0]);
parent::addRDF(parent::triplify($did, parent::getVoc() . "mixture", $mid) . parent::describeIndividual($mid, $o->name[0], parent::getVoc() . "Mixture") . parent::describeClass(parent::getVoc() . "Mixture", "mixture") . parent::triplifyString($mid, $this->getVoc() . "ingredients", "" . $o->ingredients[0]));
$a = explode(" + ", $o->ingredients[0]);
foreach ($a as $b) {
$b = trim($b);
$iid = parent::getRes() . str_replace(" ", "-", $b);
parent::addRDF(parent::describeClass($iid, $b, parent::getVoc() . "Ingredient") . parent::describeClass(parent::getVoc() . "Ingredient", "Ingredient") . parent::triplify($mid, parent::getVoc() . "ingredient", $iid));
}
}
}
}
// packagers
// <packagers><packager><name>Cardinal Health</name><url>http://www.cardinal.com</url></packager>
if (isset($x->packagers)) {
foreach ($x->packagers as $items) {
if (isset($items->packager)) {
foreach ($items->packager as $item) {
$pid = parent::getRes() . md5($item->name);
parent::addRDF(parent::triplify($did, parent::getVoc() . "packager", $pid));
if (!isset($defined[$pid])) {
$defined[$pid] = '';
parent::addRDF(parent::describe($pid, "" . $item->name[0]));
if (strstr($item->url, "http://") && $item->url != "http://BASF Corp.") {
parent::addRDF($this->triplify($pid, "rdfs:seeAlso", "" . $item->url[0]));
}
}
}
}
}
}
// manufacturers
$this->AddText($x, $did, "manufacturers", "manufacturer", parent::getVoc() . "manufacturer");
// @TODO RESOURCE
// prices
if (isset($x->prices->price)) {
foreach ($x->prices->price as $product) {
$pid = parent::getRes() . md5($product->description);
parent::addRDF(parent::describeIndividual($pid, $product->description, parent::getVoc() . "Pharmaceutical", $product->description) . parent::describeClass(parent::getVoc() . "Pharmaceutical", "pharmaceutical") . parent::triplifyString($pid, parent::getVoc() . "price", "" . $product->cost, "xsd:float") . parent::triplify($did, parent::getVoc() . "product", $pid));
$uid = parent::getVoc() . md5($product->unit);
parent::addRDF(parent::describeIndividual($uid, $product->unit, parent::getVoc() . "Unit", $product->unit) . parent::describeClass(parent::getVoc() . "Unit", "unit") . parent::triplify($pid, parent::getVoc() . "form", $uid));
}
}
// dosages <dosages><dosage><form>Powder, for solution</form><route>Intravenous</route><strength></strength></dosage>
if (isset($x->dosages->dosage)) {
foreach ($x->dosages->dosage as $dosage) {
$id = parent::getRes() . md5($dosage->strength . $dosage->form . $dosage->route);
$label = ($dosage->strength != '' ? $dosage->strength . " " : "") . $dosage->form . " form with " . $dosage->route . " route";
parent::addRDF(parent::describeIndividual($id, $label, parent::getVoc() . "Dosage") . parent::describeClass(parent::getVoc() . "Dosage", "Dosage") . parent::triplify($did, parent::getVoc() . "dosage", $id));
$rid = parent::getVoc() . md5($dosage->route);
$this->typify($id, $rid, "Route", "" . $dosage->route);
$fid = parent::getVoc() . md5($dosage->form);
$this->typify($id, $fid, "Form", "" . $dosage->form);
if ($dosage->strength != '') {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "strength", $dosage->strength));
}
}
}
// experimental-properties
$props = array("experimental-properties", "calculated-properties");
foreach ($props as $prop) {
$subtype = substr($prop, 0, strpos("-", $prop));
if (isset($x->{$prop})) {
foreach ($x->{$prop} as $properties) {
foreach ($properties as $property) {
$type = (string) $property->kind;
$value = (string) $property->value;
$type_uri = parent::getVoc() . ucfirst(str_replace(" ", "-", $type));
$id = parent::getRes() . $prop . "-" . $dbid . "-" . $counter++;
$label = $property->kind . ": {$value}" . ($property->source == '' ? '' : " from " . $property->source);
parent::addRDF(parent::describeIndividual($id, $label, $type_uri) . parent::describeClass($type_uri, $type, parent::getVoc() . ucfirst($prop)) . parent::describeClass(parent::getVoc() . ucfirst($prop), str_replace("-", " ", $prop)) . parent::triplifyString($id, $this->getVoc() . "value", $value) . parent::triplify($did, $this->getVoc() . $prop, $id));
// Source
if (isset($property->source)) {
foreach ($property->source as $source) {
$s = (string) $source;
if ($s == '') {
continue;
}
$sid = parent::getRes() . md5($s);
parent::addRDF(parent::describeIndividual($sid, $s, parent::getVoc() . "Source") . parent::describeClass(parent::getVoc() . "Source", "Source") . parent::triplify($id, parent::getVoc() . "source", $sid));
}
}
}
}
}
}
// identifiers
// <patents><patent><number>RE40183</number><country>United States</country><approved>1996-04-09</approved> <expires>2016-04-09</expires>
if (isset($x->patents->patent)) {
foreach ($x->patents->patent as $patent) {
$id = "uspto:" . $patent->number;
parent::addRDF(parent::triplify($did, $this->getVoc() . "patent", $id) . parent::describeIndividual($id, $patent->country . " patent " . $patent->number, $this->getVoc() . "Patent") . parent::describeClass(parent::getVoc() . "Patent", "patent") . parent::triplifyString($id, $this->getVoc() . "approved", "" . $patent->approved) . parent::triplifyString($id, $this->getVoc() . "expires", "" . $patent->expires));
$cid = parent::getRes() . md5($patent->country);
$this->typify($id, $cid, "Country", "" . $patent->country);
}
}
// partners
$partners = array('target', 'enzyme', 'transporter', 'carrier');
foreach ($partners as $partner) {
$plural = $partner . 's';
if (isset($x->{$plural})) {
foreach ($x->{$plural} as $list) {
foreach ($list->{$partner} as $item) {
$this->parsePartnerRelation($did, $item, $partner);
parent::writeRDFBufferToWriteFile();
}
}
}
}
// drug-interactions
$y = (int) substr($dbid, 2);
if (isset($x->{"drug-interactions"})) {
foreach ($x->{"drug-interactions"} as $ddis) {
foreach ($ddis->{"drug-interaction"} as $ddi) {
$dbid2 = $ddi->{'drugbank-id'};
if ($dbid < $dbid2) {
// don't repeat
$ddi_id = parent::getRes() . $dbid . "_" . $dbid2;
parent::addRDF(parent::triplify("drugbank:" . $dbid, parent::getVoc() . "ddi-interactor-in", "" . $ddi_id) . parent::triplify("drugbank:" . $dbid2, parent::getVoc() . "ddi-interactor-in", "" . $ddi_id) . parent::describeIndividual($ddi_id, "DDI between {$name} and " . $ddi->name . " - " . $ddi->description, parent::getVoc() . "Drug-Drug-Interaction") . parent::describeClass(parent::getVoc() . "Drug-Drug-Interaction", "drug-drug interaction"));
}
}
}
}
// food-interactions
$this->AddText($x, $did, "food-interactions", "food-interaction", parent::getVoc() . "food-interaction");
// affected-organisms
$this->AddCategory($x, $did, "affected-organisms", "affected-organism", parent::getVoc() . "affected-organism");
// <external-identifiers>
if (isset($x->{"external-identifiers"})) {
foreach ($x->{"external-identifiers"} as $objs) {
foreach ($objs as $obj) {
$ns = $this->NSMap($obj->resource);
$id = $obj->identifier;
if ($ns == "genecards") {
$id = str_replace(array(" "), array("_"), $id);
}
parent::addRDF(parent::triplify($did, parent::getVoc() . "x-{$ns}", "{$ns}:{$id}"));
if ($ns == "pubchemcompound") {
parent::addRDF(parent::triplify("{$ns}:{$id}", "skos:exactMatch", "http://rdf.ncbi.nlm.nih.gov/pubchem/compound/{$id}"));
}
}
}
}
// <external-links>
if (isset($x->{"external-links"})) {
foreach ($x->{"external-links"}->{'external-link'} as $el) {
if (strpos($el->url, 'http') !== false) {
parent::addRDF(parent::triplify($did, "rdfs:seeAlso", "" . $el->url));
}
}
}
parent::writeRDFBufferToWriteFile();
}
function genes($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 freq()
{
$cols = 10;
$i = 1;
parent::setCheckpoint('file');
while ($l = parent::getReadFile()->read()) {
$a = explode("\t", str_replace("%", "", $l));
if (count($a) != $cols) {
trigger_error("Expecting {$cols}, but found " . count($a) . " instead... skipping file!", E_USER_ERROR);
return false;
}
list($stitch_flat, $stitch_stereo, $cui, $placebo, $freq, $freq_lower, $freq_upper, $concept_type, $meddra_concept_id, $meddra_concept_label) = $a;
if ($concept_type == "LLT") {
continue;
}
$meddra_concept_label = trim($meddra_concept_label);
$id = "stitch_resource:" . md5("se_freq" . $l);
$stitch_flat = "stitch:{$stitch_flat}";
$label = "{$meddra_concept_label} frequency for {$stitch_flat}";
parent::addRDF(parent::describeIndividual($id, $label, parent::getVoc() . "Drug-Effect-Frequency") . parent::describeClass(parent::getVoc() . "Drug-Effect-Frequency", "SIDER Drug-Effect and Frequency") . parent::triplify($id, parent::getVoc() . "drug", $stitch_flat) . parent::triplify($id, parent::getVoc() . "effect", "umls:" . $meddra_concept_id));
if ($placebo) {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "placebo", "true", "xsd:boolean"));
}
$number = false;
if (is_numeric($freq)) {
$flabel = $freq . "%";
$ftype_label = "Exact-Frequency";
$ftype = parent::getVoc() . $ftype_label;
$number = true;
} else {
$flabel = $freq;
$ftype_label = "Qualitative-Frequency";
$ftype = parent::getVoc() . "{$ftype_label}";
}
if ($freq_lower != $freq_upper) {
$flabel .= "({$freq_lower}-{$freq_upper})";
$ftype_label = "Range-Frequency";
$ftype = parent::getVoc() . $ftype_label;
}
$fid = $id . md5($a[5] . $a[6] . $a[8]);
parent::addRDF(parent::triplify($id, parent::getVoc() . "frequency", $fid) . parent::describeIndividual($fid, $flabel, $ftype) . parent::describeClass($ftype, $ftype_label));
if ($number == true) {
parent::addRDF(parent::triplifyString($fid, parent::getVoc() . "frequency-value", $freq / 100));
} else {
parent::addRDF(parent::triplifyString($fid, parent::getVoc() . "frequency-value", $freq));
}
parent::addRDF(parent::triplifyString($fid, parent::getVoc() . "lower-frequency", sprintf("%.3f", $freq_lower)) . parent::triplifyString($fid, parent::getVoc() . "upper-frequency", sprintf("%.3f", $freq_upper)));
parent::setCheckpoint('record');
}
parent::setCheckpoint('file');
}
function addDate($id, $field, $dateobj)
{
if ($dateobj == null) {
return FALSE;
}
$year = $dateobj->Year;
$month = $dateobj->Month;
$day = $dateobj->Day;
parent::addRDF(parent::triplifyString($id, parent::getVoc() . $field, "{$year}-{$month}-{$day}", "xsd:date"));
}
function parseKGML($lfile)
{
$pathway = simplexml_load_file($lfile);
if ($pathway === false) {
echo "Error in parsing {$lfile}" . PHP_EOL;
return;
}
$pathway_id = str_replace("path", "kegg", $pathway['name']);
$base_id = str_replace("kegg", "kegg_resource", $pathway_id) . ".";
parent::addRDF(parent::describeIndividual($pathway_id, $pathway['title'], parent::getVoc() . "Pathway") . parent::triplify($pathway_id, "rdfs:seeAlso", $pathway['link']) . parent::triplify($pathway_id, "foaf:depiction", $pathway['image']));
// get the entries
foreach ($pathway->children() as $type => $item) {
if ($type == "entry") {
$eid = $base_id . $item['id'];
$entries["" . $item['id']] = "" . $item['name'];
parent::addRDF(parent::describeIndividual($eid, $item['name'], parent::getVoc() . "Ortholog-Group") . parent::describeClass(parent::getVoc() . "Ortholog-Group", "KEGG Ortholog Group"));
$mids = explode(" ", $item['name']);
foreach ($mids as $mid) {
if ($item['type'] == 'path') {
$mid = str_replace($mid, ":", "_");
} else {
$mid = substr($mid, strpos($mid, ":") + 1);
}
parent::addRDF(parent::triplify($eid, parent::getVoc() . "member", "kegg:" . $mid));
}
}
}
// iterate over the relations, reactions
foreach ($pathway->children() as $type => $item) {
if ($type == "relation") {
/*
<relation entry1="70" entry2="73" type="ECrel">
<subtype name="compound" value="86"/>
</relation>
<relation entry1="26" entry2="25" type="PPrel">
<subtype name="compound" value="17"/>
<subtype name="activation" value="-->"/>
</relation>
*/
$id1 = "" . $item['entry1'];
$id2 = "" . $item['entry2'];
$type = "" . $type;
$relation_id = str_replace("kegg", "kegg_resource", $pathway_id) . "." . $id1 . "." . $id2 . "." . $type;
$label = $type . " relation between " . $entries[$id1] . " and " . $entries[$id2];
parent::addRDF(parent::describeIndividual($relation_id, $label, parent::getVoc() . "Pathway-Relation") . parent::describeClass(parent::getVoc() . "Pathway-Relation", "KEGG Pathway Relation") . parent::triplify($relation_id, parent::getVoc() . "source", $base_id . $id1) . parent::triplify($relation_id, parent::getVoc() . "target", $base_id . $id2) . parent::triplify($relation_id, parent::getVoc() . "pathway", $pathway_id) . parent::triplifyString($relation_id, parent::getVoc() . "type", $item['type']));
foreach ($item->children() as $subtype) {
parent::addRDF(parent::triplifyString($relation_id, parent::getVoc() . "subtype", '' . $subtype['name']));
}
} else {
if ($type == "reaction") {
/* <reaction id="133" name="rn:R09085" type="irreversible">
<substrate id="86" name="cpd:C00267"/>
<product id="90" name="cpd:C00668"/>
</reaction>
*/
$reaction_id = str_replace("kegg", "kegg_resource", $pathway_id) . "." . substr($item['name'], strpos($item['name'], ":") + 1);
$reaction_type = parent::getVoc() . ucfirst($item['type']) . "-Reaction";
parent::addRDF(parent::describeIndividual($reaction_id, $item['name'], parent::getVoc() . "Reaction") . parent::describeClass(parent::getVoc() . "Reaction", "KEGG Reaction") . parent::triplify($reaction_id, "rdf:type", $reaction_type));
foreach ($item->children() as $k => $v) {
$cid = str_replace("cpd:", "kegg:", $v['name']);
parent::addRDF(parent::triplify($reaction_id, parent::getVoc() . $k, $cid));
}
}
}
}
return;
}
private function process()
{
//read the file into a string
$xml_str = '';
while ($l = $this->getReadFile()->Read(4096)) {
$xml_str .= $l;
}
$xml = new SimpleXMLElement($xml_str);
foreach ($xml->{'sts'} as $sts) {
$uid = (string) $sts->uid;
$uid_res = $this->getNamespace() . $uid;
$name = (string) $sts->name;
$uid_label = $name;
parent::AddRDF(parent::describeIndividual($uid_res, $uid_label, $this->getVoc() . "sequence_tagged_site"));
//split by semicolon
$genbank_accession_arr = explode(";", (string) $sts->gbacc);
foreach ($genbank_accession_arr as $gb) {
parent::AddRDF(parent::triplify($uid_res, $this->getVoc() . 'x-genbank', 'genbank:' . $gb));
}
//split by semicolon
$gi_arr = explode(';', (string) $sts->gi);
foreach ($gi_arr as $gi) {
parent::AddRDF(parent::triplify($uid_res, $this->getVoc() . 'x-gi', 'gi:' . $gi));
}
$links = (string) $sts->links;
if (isset($links)) {
parent::AddRDF(parent::triplifyString($uid_res, $this->getVoc() . 'links', utf8_encode($links)));
}
$forward_primer = (string) $sts->pcrfor;
if (isset($forward_primer)) {
parent::AddRDF(parent::triplifyString($uid_res, $this->getVoc() . 'forward-primer', $forward_primer));
}
$reverse_primer = (string) $sts->pcrrev;
if (isset($reverse_primer)) {
parent::AddRDF(parent::triplifyString($uid_res, $this->getVoc() . 'reverse-primer', $reverse_primer));
}
$epcr_summary = (string) $sts->EPCR_Summary;
if (isset($epcr_summary)) {
parent::AddRDF(parent::triplifyString($uid_res, $this->getVoc() . 'epcr-summary', $epcr_summary));
}
$dbsts = (string) $sts->dbsts;
if (isset($dbsts)) {
parent::AddRDF(parent::triplifyString($uid_res, $this->getVoc() . 'dbsts', $dbsts));
}
foreach ($sts->{'Map_Gene_Summary_List'} as $gsl) {
foreach ($gsl->{'Map_Gene_Summary'} as $mgs) {
$r = rand();
$org = (string) $mgs->Org;
$taxid = (string) $mgs->taxid;
$chromosome = (string) $mgs->Chr;
$locus = (string) $mgs->Locus;
$polymorphic = (string) $mgs->Polymorphic;
$lbl = "map gene summary for :" . $org . " " . $taxid;
$u_res = $this->getRes() . md5($r . $org);
parent::AddRDF(parent::describeIndividual($u_res, $lbl, $this->getVoc() . "map-gene-summary") . parent::triplify($uid_res, $this->getVoc() . "has-map-gene-summary", $u_res));
if (isset($polymorphic)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "polymorphic", $polymorphic));
}
if (isset($locus)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "locus", $locus));
}
if (isset($chromosome)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "chromosome", $chromosome));
}
if (isset($taxid)) {
parent::AddRDF(parent::triplify($u_res, $this->getVoc() . "x-taxonomy", "taxon:" . $taxid));
}
if (isset($org)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "org-name", $org));
}
}
}
foreach ($sts->{'org'} as $anOrg) {
$r = rand();
$taxname = (string) $anOrg->taxname;
$taxid = (string) $anOrg->taxid;
$name = (string) $anOrg->name;
$keyword = (string) $anOrg->keyword;
$alias = (string) $anOrg->alias;
$polymorphic = (string) $anOrg->polymorphic;
$dseg = (string) $anOrg->dseg;
$gdb = (string) $anOrg->gdb;
$rgd = (string) $anOrg->rgd;
$mgd = (string) $anOrg->mgd;
$zfin = (string) $anOrg->zfin;
$pcrsize = (string) $anOrg->pcrsize;
$dbsnp = (string) $anOrg->dbsnp;
$warn = (string) $anOrg->warn;
$wloc = (string) $anOrg->wloc;
$wmap = (string) $anOrg->wmap;
$wctg = (string) $anOrg->wctg;
$wdrift = (string) $anOrg->wdrift;
$mapview = (string) $anOrg->mapview;
$u_res = $this->getRes() . md5($r . $taxname . $taxid);
$u_lbl = "org entry for " . $taxname . " " . $taxid;
parent::AddRDF(parent::describeIndividual($u_res, $u_lbl, $this->getVoc() . "organism") . parent::triplify($uid_res, $this->getVoc() . "has-organism", $u_res));
if (isset($mapview) && strlen($mapview)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "mapview", $mapview));
}
if (isset($wdrift) && strlen($wdrift)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "wdrift", $wdrift));
}
if (isset($wctg) && strlen($wctg)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "wctg", $wctg));
}
if (isset($wmap) && strlen($wmap)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "wmap", $wmap));
}
if (isset($wloc) && strlen($wloc)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "wloc", $wloc));
}
if (isset($warn) && strlen($warn)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "warn", $warn));
}
if (isset($taxname) && strlen($taxname)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "taxname", $taxname));
}
if (isset($name)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "name", $name));
}
if (isset($keyword) && strlen($keyword)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "keyword", $keyword));
}
if (isset($alias) && strlen($alias)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "alias", $alias));
}
if (isset($polymorphic) && strlen($polymorphic)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "polymorphic", $polymorphic));
}
if (isset($dseg) && strlen($dseg)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "dseg", $dseg));
}
if (isset($gdb) && strlen($gdb)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "gdb", $gdb));
}
if (isset($rgd) && strlen($rgd)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "rgd", $rgd));
}
if (isset($mgd) && strlen($mgd)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "mgd", $mgd));
}
if (isset($zfin) && strlen($zfin)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "zfin", $zfin));
}
if (isset($pcrsize) && strlen($pcrsize)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "pcrsize", $pcrsize));
}
if (isset($dbsnp) && strlen($dbsnp)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . "dbsnp", $dbsnp));
}
foreach ($anOrg->{'rhdb'} as $rh) {
$id = (string) $rh->id;
$panel = (string) $rh->panel;
$r = rand();
$u_res = $this->getRes() . md5($r . $id . $panel);
$u_label = "rhdb for " . $id . " " . $panel;
parent::AddRDF(parent::describeIndividual($u_res, $u_label, $this->getVoc() . "rhdb") . parent::triplify($uid_res, $this->getVoc() . 'has-rhdb', $u_res) . parent::triplifyString($u_res, $this->getVoc() . 'id', $id) . parent::triplifyString($u_res, $this->getVoc() . 'panel', $panel));
}
foreach ($anOrg->{'unigene'} as $u) {
$ugid = (string) $u->unigene;
$ugname = (string) $u->ugname;
$r = rand();
$u_res = $this->getRes() . md5($r . $ugid . $ugname);
$u_label = "unigene for " . $ugid . " " . $ugname;
if (strlen($ugid) > 0 && strlen($ugname) > 0) {
parent::AddRDF(parent::describeIndividual($u_res, $u_label, $this->getVoc() . "unigene") . parent::triplify($uid_res, $this->getVoc() . 'has-unigene', $u_res) . parent::triplify($u_res, $this->getVoc() . 'x-unigene', $ugid) . parent::triplifyString($u_res, $this->getVoc() . 'ugname', $ugname));
}
}
foreach ($anOrg->{'locus'} as $aLocus) {
$lid = (string) $aLocus->lid;
$lsymbol = (string) $aLocus->lsymbol;
$lname = (string) $aLocus->lname;
$lcyto = (string) $aLocus->lcyto;
$ltype = (string) $aLocus->ltype;
$r = rand();
$u_res = $this->getRes() . md5($r . $lid);
$u_label = "locus for " . $lid . " " . $lname . " " . $lsymbol;
parent::AddRDF(parent::describeIndividual($u_res, $u_label, $this->getVoc() . "locus") . parent::triplify($uid_res, $this->getVoc() . 'has-locus', $u_res));
if (isset($lcyto) && strlen($lcyto)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'cyto', $lcyto));
}
if (isset($ltype) && strlen($ltype)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'ltype', $ltype));
}
if (isset($lname) && strlen($lname)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'name', $lname));
}
}
foreach ($anOrg->{'mappos'} as $aMapPos) {
$map = (string) $aMapPos->map;
$marker = (string) $aMapPos->marker;
$chr = (string) $aMapPos->chr;
$chrpos = (string) $aMapPos->chrpos;
$ctg = (string) $aMapPos->ctg;
$ctgpos = (string) $aMapPos->ctgpos;
$coord = (string) $aMapPos->coord;
$lod = (string) $aMapPos->lod;
$bin = (string) $aMapPos->bin;
$binpos = (string) $aMapPos->binpos;
$het = (string) $aMapPos->het;
$lab = (string) $aMapPos->lab;
$unit = (string) $aMapPos->unit;
$r = rand();
$u_res = $this->getRes() . md5($r . $map);
$u_label = "locus for " . $map . " " . $marker . " " . $chr;
parent::AddRDF(parent::describeIndividual($u_res, $u_label, $this->getVoc() . "mappos") . parent::triplify($uid_res, $this->getVoc() . 'has-mappos', $u_res));
if (isset($unit) && strlen($unit)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'unit', $unit));
}
if (isset($lab) && strlen($lab)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'lab', $lab));
}
if (isset($het) && strlen($het)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'het', $het));
}
if (isset($binpos) && strlen($binpos)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'binpos', $binpos));
}
if (isset($bin) && strlen($bin)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'bin', $bin));
}
if (isset($lod) && strlen($lod)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'lod', $lod));
}
if (isset($coord) && strlen($coord)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'coord', $coord));
}
if (isset($ctgpos) && strlen($ctgpos)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'ctgpos', $ctgpos));
}
if (isset($ctg) && strlen($ctg)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'ctg', $ctg));
}
if (isset($chrpos) && strlen($chrpos)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'chrpos', $chrpos));
}
if (isset($chr) && strlen($chr)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'chr', $chr));
}
if (isset($marker) && strlen($marker)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'marker', $marker));
}
}
foreach ($anOrg->{'epcr'} as $aEpcr) {
$seqType = (string) $aEpcr->seqtype;
$acc = (string) $aEpcr->acc;
$gi = (string) $aEpcr->gi;
$pos1 = (string) $aEpcr->pos1;
$pos2 = (string) $aEpcr->pos2;
$epcr_size = (string) $aEpcr->epcrsize;
$r = rand();
$u_res = $this->getRes() . md5($r . $gi . $pos1);
$u_label = "epcr for " . $pos1 . " " . $pos2;
parent::AddRDF(parent::describeIndividual($u_res, $u_label, $this->getVoc() . "epcr") . parent::triplify($uid_res, $this->getVoc() . 'has-epcr', $u_res));
if (isset($seqtype) && strlen($seqtype)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'seqtype', $seqtype));
}
if (isset($acc) && strlen($acc)) {
parent::AddRDF(parent::triplify($u_res, $this->getVoc() . 'x-genbank', 'genbank:' . $acc));
}
if (isset($gi) && strlen($gi)) {
parent::AddRDF(parent::triplify($u_res, $this->getVoc() . 'x-gi', 'gi:' . $gi));
}
if (isset($pos1) && strlen($pos1)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'pos1', $pos1));
}
if (isset($pos2) && strlen($pos2)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'pos2', $pos2));
}
if (isset($epcr_size) && strlen($epcr_size)) {
parent::AddRDF(parent::triplifyString($u_res, $this->getVoc() . 'epcr_size', $epcr_size));
}
}
}
$this->WriteRDFBufferToWriteFile();
}
}
function MGI_Geno_NotDisease()
{
$line = 1;
while ($l = $this->getReadFile()->read(248000)) {
$a = explode("\t", $l);
if (count($a) != 8) {
trigger_error("Incorrect number of columns", E_USER_WARNING);
continue;
}
$genotype = $a[0];
$alleles = explode("|", strtolower($a[2]));
$diseases = explode(",", $a[7]);
foreach ($diseases as $d) {
$disease = "omim:{$d}";
foreach ($alleles as $allele) {
$id = parent::getRes() . md5($allele . $disease);
$label = "{$allele} {$disease} absent association";
parent::addRDF(parent::describeIndividual($id, $label, $this->getVoc() . "Allele-Disease-Non-Association") . parent::describeClass($this->getVoc() . "Allele-Disease-Non-Association", "MGI Allele-Disease Non-Association") . parent::triplify($id, $this->getVoc() . "allele", $allele) . parent::triplifyString($id, $this->getVoc() . "genotype-string", $genotype) . parent::triplify($id, $this->getVoc() . "disease", $disease) . parent::triplifyString($id, $this->getVoc() . "is-negated", "true"));
if ($a[5]) {
$pmids = explode(",", $a[5]);
foreach ($pmids as $pmid) {
parent::addRDF(parent::triplify($id, $this->getVoc() . "x-pubmed", "pubmed:" . $pmid));
}
}
}
}
$this->writeRDFBufferToWriteFile();
}
}
function Parse($file)
{
parent::getReadFile()->read();
// skip the first comment line
$line = 1;
$first = true;
while ($l = parent::getReadFile()->read(500000)) {
if ($l[0] == "#") {
// dataset attributes
$a = explode('=', trim($l));
$r = $this->getVoc() . substr($a[0], 2);
if (isset($a[1])) {
$v = $a[1];
if ($r == "affymetrix_vocabulary:genome-version-create_date") {
$x = explode("-", $a[1]);
if ($x[2] == "00") {
$x[2] = "01";
}
$v = implode("-", $x);
}
parent::addRDF(parent::triplifyString(parent::getDatasetURI(), $r, $v) . parent::describe($r, "{$r}"));
}
continue;
}
if ($first == true) {
$first = false;
// header
$header = explode(",", str_replace('"', '', trim($l)));
// print_r($header);exit;
$n = count($header);
if ($n != 41) {
trigger_error("Expecting 41 columns, found {$n} in header on line {$line}!", E_USER_ERROR);
exit;
}
continue;
}
$a = explode('","', substr($l, 1, -2));
$n = count($a);
if ($n != 41) {
trigger_error("Expecting 41 columns, found {$n} on line {$line}!", E_USER_ERROR);
exit;
}
parent::writeRDFBufferToWriteFile();
$id = $a[0];
$qname = "affymetrix:{$id}";
$label = "probeset {$a['0']} on GeneChip {$a['1']} ({$a['2']})";
parent::addRDF(parent::describeIndividual($qname, $label, $this->getVoc() . "Probeset") . parent::describeClass($this->getVoc() . "Probeset", "Affymetrix probeset"));
trigger_error($id, E_USER_NOTICE);
// now process the entries
foreach ($a as $k => $v) {
if (trim($v) == '---') {
continue;
}
// multi-valued entries are separated by ////
$b = explode(" /// ", $v);
$r = $this->Map($k);
if (isset($r)) {
foreach ($b as $c) {
$d = explode(" // ", $c);
if ($r == 'symbol') {
$d[0] = str_replace(" ", "-", $d[0]);
}
$s = $this->getRegistry()->getPreferredPrefix($r);
if ($s == "ec") {
$e = explode(":", $d[0]);
$d[0] = $e[1];
}
$this->addRDF(parent::triplify($qname, $this->getVoc() . "x-{$s}", "{$s}:" . $d[0]) . parent::describeProperty($this->getVoc() . "x-{$s}", "a relation to {$s}"));
}
} else {
// we handle manually
unset($rel);
$label = $header[$k];
switch ($label) {
case 'GeneChip Array':
$array_id = parent::getRes() . str_replace(" ", "-", $v);
parent::addRDF(parent::triplify($qname, $this->getVoc() . "genechip-array", $array_id) . parent::describeIndividual($array_id, "Affymetrix {$v} GeneChip array", $this->getVoc() . "Genechip-Array") . parent::describeClass($this->getVoc() . "Genechip-Array", "Affymetrix GeneChip array"));
break;
case 'Gene Ontology Biological Process':
if (!isset($rel)) {
$rel = 'go-process';
$prefix = "go";
}
case 'Gene Ontology Cellular Component':
if (!isset($rel)) {
$rel = 'go-location';
$prefix = "go";
}
case 'Gene Ontology Molecular Function':
if (!isset($rel)) {
$rel = 'go-function';
$prefix = "go";
}
$b = explode(" /// ", $v);
foreach ($b as $c) {
$d = explode(" // ", $c);
parent::addRDF($this->triplify($qname, $this->getVoc() . $rel, "{$prefix}:" . $d[0]) . $this->describeProperty($this->getVoc() . $rel, "{$rel}"));
}
break;
case 'Transcript Assignments':
$b = explode(" /// ", $v);
foreach ($b as $c) {
$d = explode(" // ", $c);
$id = $d[0];
$prefix = $d[2];
if ($prefix == '---' || $id == '---') {
continue;
} else {
if ($prefix == 'gb' || $prefix == 'gb_htc') {
$prefix = 'genbank';
} else {
if ($prefix == 'ncbibacterial') {
$prefix = 'gi';
} else {
if ($prefix == 'ncbi_bacterial') {
$prefix = 'gi';
} else {
if ($prefix == 'ens') {
$prefix = 'ensembl';
} else {
if ($prefix == 'ncbi_mito' || $prefix == 'ncbi_organelle' || $prefix == 'organelle') {
$prefix = 'refseq';
} else {
if ($prefix == 'affx' || $prefix == 'unknown' || $prefix == "prop") {
$prefix = 'affymetrix';
} else {
if ($prefix == 'tigr_2004_08') {
$prefix = 'tigr';
} else {
if ($prefix == 'tigr-plantta') {
$prefix = 'genbank';
} else {
if ($prefix == 'newrs.gi') {
$prefix = 'gi';
} else {
if ($prefix == 'newRS.gi') {
$prefix = 'gi';
} else {
if ($prefix == 'primate_viral') {
$prefix = 'genbank';
} else {
if ($prefix == 'jgi-bacterial') {
$prefix = 'ncbigene';
} else {
if ($prefix == 'tb') {
$prefix = 'tuberculist';
} else {
if ($prefix == 'pa') {
$prefix = 'pseudomonas';
} else {
if ($prefix == 'gi|53267') {
$prefix = 'gi';
$id = '53267';
} else {
if ($prefix == 'broad-tcup') {
$e = explode("-", $id);
$id = $e[0];
} else {
if ($prefix == 'organelle') {
$e = explode("-", $id);
$prefix = 'genbank';
$id = $e[0];
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
}
parent::addRDF(parent::triplify($qname, $this->getVoc() . "transcript-assignment", "{$prefix}:{$id}") . parent::describeProperty($this->getVoc() . "transcript-assignment", "transcript assignment"));
}
break;
case 'Annotation Transcript Cluster':
/*
$id = substr($v,0,strpos($v,"("));
$rel = str_replace(" ","-",strtolower($label));
$this->AddRDF($this->triplify($qname,parent::getVoc()."$rel", "refseq:$id"));
*/
break;
case 'Annotation Date':
// Jun 9, 2011
$rel = "annotation-date";
preg_match("/^([A-Za-z]+) ([0-9]+), ([0-9]{4})\$/", $v, $m);
if (count($m) == 4) {
array_shift($m);
list($m, $day, $year) = $m;
$month = $this->getMonth($m);
if (!$day || $day == "0") {
$day = "01";
}
$date = $year . "-" . $month . "-" . str_pad($day, 2, "0", STR_PAD_LEFT) . "T00:00:00Z";
parent::addRDF(parent::triplifyString($qname, $this->getVoc() . $rel, $date, "xsd:dateTime") . parent::describeProperty($this->getVoc() . $rel, "{$rel}"));
} else {
trigger_error("could not match date from {$v}", E_USER_ERROR);
}
break;
case 'Species Scientific Name':
break;
case 'Transcript ID(Array Design)':
if (!isset($rel)) {
$rel = 'transcript';
}
case 'Sequence type':
default:
if (!isset($rel)) {
$rel = str_replace(" ", "-", strtolower($label));
}
$b = explode(" /// ", $v);
foreach ($b as $c) {
parent::addRDF(parent::triplifyString($qname, $this->getVoc() . $rel, stripslashes($c)) . parent::describeProperty($this->getVoc() . $rel, "{$rel}"));
}
break;
}
// switch
}
// else
}
$this->WriteRDFBufferToWriteFile();
}
}
function process()
{
while ($aLine = $this->GetReadFile()->Read(200000)) {
$parsed_line = $this->parse_homologene_tab_line($aLine);
$hid = "homologene:" . $parsed_line["hid"];
$hid_label = "homologene group " . $parsed_line['hid'];
parent::AddRDF(parent::describeIndividual($hid, $hid_label, $this->getVoc() . "Homologene-Group") . parent::describeClass($this->getVoc() . "Homologene-Group", "Homologene Group"));
$geneid = "ncbigene:" . $parsed_line["geneid"];
$taxid = "taxid:" . $parsed_line["taxid"];
$gi = "gi:" . $parsed_line["gi"];
$genesymbol = str_replace("\\", "", $parsed_line["genesymbol"]);
$refseq = "refseq:" . $parsed_line["refseq"];
parent::AddRDF(parent::triplify($hid, $this->getVoc() . "x-taxid", $taxid) . parent::describeProperty($this->getVoc() . "x-taxid", "Link to NCBI taxonomy"));
parent::AddRDF(parent::triplify($hid, $this->getVoc() . "x-ncbigene", $geneid) . parent::describeProperty($this->getVoc() . "x-ncbigene", "Link to NCBI GeneId"));
parent::AddRDF(parent::triplifyString($hid, $this->getVoc() . "gene-symbol", utf8_encode(htmlspecialchars($genesymbol)), "xsd:string") . parent::describeProperty($this->getVoc() . "gene-symbol", "Link to gene symbol"));
parent::AddRDF(parent::triplify($hid, $this->getVoc() . "x-gi", $gi) . parent::describeProperty($this->getVoc() . "x-gi", "Link to NCBI GI"));
parent::AddRDF(parent::triplify($hid, $this->getVoc() . "x-refseq", $refseq) . parent::describeProperty($this->getVoc() . "x-refseq", "Link to NCBI Refseq"));
$this->WriteRDFBufferToWriteFile();
}
}
function process()
{
$header = $this->GetReadFile()->Read(200000);
$header_arr = explode("\t", $header);
$n = 41;
$c = count($header_arr);
if ($c != $n) {
echo PHP_EOL;
print_r($header_arr);
trigger_error("Expected {$n} columns, found {$c} . please update the script", E_USER_ERROR);
exit;
}
while ($l = $this->GetReadFile()->Read(4096)) {
$fields = explode("\t", $l);
$id = strtolower($fields[0]);
$approved_symbol = $fields[1];
$approved_name = $fields[2];
$status = $fields[3];
$locus_type = $fields[4];
$locus_group = $fields[5];
$previous_symbols = $fields[6];
$previous_names = $fields[7];
$synonyms = $fields[8];
$name_synonyms = $fields[9];
$chromosome = $fields[10];
$date_approved = $fields[11];
$date_modified = $fields[12];
$date_symbol_changed = $fields[13];
$date_name_changed = $fields[14];
$accession_numbers = $fields[15];
$enzyme_ids = $fields[16];
$entrez_gene_id = $fields[17];
$ensembl_gene_id = $fields[18];
$mouse_genome_database_id = $fields[19];
$specialist_database_links = $fields[20];
$specialist_database_ids = $fields[21];
$pubmed_ids = $fields[22];
$refseq_ids = $fields[23];
$gene_family_tag = $fields[24];
$gene_family_description = $fields[25];
$record_type = $fields[26];
$primary_ids = $fields[27];
$secondary_ids = $fields[28];
$ccd_ids = $fields[29];
$vega_ids = $fields[30];
$locus_specific_databases = $fields[31];
$entrez_gene_id_mappeddatasuppliedbyNCBI = $fields[32];
$omim_id_mappeddatasuppliedbyNCBI = $fields[33];
$refseq_mappeddatasuppliedbyNCBI = $fields[34];
$uniprot_id_mappeddatasuppliedbyUniProt = $fields[35];
$ensembl_id_mappeddatasuppliedbyEnsembl = $fields[36];
$vega_id_mappeddatasuppliedbyVega = $fields[37];
$ucsc_id_mappeddatasuppliedbyUCSC = $fields[38];
$mouse_genome_database_id_mappeddatasuppliedbyMGI = $fields[39];
$rat_genome_database_id_mappeddatasuppliedbyRGD = $fields[40];
$id_res = $id;
$id_label = "Gene Symbol for " . $approved_symbol;
parent::AddRDF(parent::triplify($id_res, "rdf:type", $this->getVoc() . "Gene-Symbol") . parent::describeIndividual($id_res, $id_label, $this->getVoc() . "Gene-Symbol") . parent::describeClass($this->getVoc() . "Gene-Symbol", "HGNC Official Gene Symbol"));
if (!empty($approved_symbol)) {
$s = "hgnc.symbol:" . $approved_symbol;
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "approved-symbol", utf8_encode(htmlspecialchars($approved_symbol))) . parent::describeProperty($this->getVoc() . "approved-symbol", "HGNC approved gene symbol", "The official gene symbol that has been approved by the HGNC and is publicly available. Symbols are approved based on specific HGNC nomenclature guidelines. In the HTML results page this ID links to the HGNC Symbol Report for that gene") . parent::describeIndividual($s, $approved_symbol, parent::getVoc() . "Approved-Gene-Symbol") . parent::describeClass(parent::getVoc() . "Approved-Gene-Symbol", "Approved Gene Symbol") . parent::triplify($id_res, parent::getVoc() . "has-approved-symbol", $s) . parent::triplify($s, parent::getVoc() . "is-approved-symbol-of", $id_res));
}
if (!empty($approved_name)) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "approved-name", utf8_encode(htmlspecialchars($approved_name))) . parent::describeProperty($this->getVoc() . "approved-name", "HGNC approved name", "The official gene name that has been approved by the HGNC and is publicly available. Names are approved based on specific HGNC nomenclature guidelines."));
}
if (!empty($status)) {
$s = $this->getVoc() . str_replace(" ", "-", $status);
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "status", $s) . parent::describeProperty($this->getVoc() . "status", "HGNC status", "Indicates whether the gene is classified as: Approved - these genes have HGNC-approved gene symbols. Entry withdrawn - these previously approved genes are no longer thought to exist. Symbol withdrawn - a previously approved record that has since been merged into a another record.") . parent::describeClass($s, $status, $this->getVoc() . "Status"));
}
if (!empty($locus_id)) {
$locus_res = $this->getRes() . $id . "_LOCUS";
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "locus", $locus_res) . parent::triplifyString($locus_res, $this->getVoc() . "locus-type", utf8_encode(htmlspecialchars($locus_type))) . parent::triplifyString($locus_res, $this->getVoc() . "locus-group", utf8_encode(htmlspecialchars($locus_group))) . parent::describeProperty($this->getVoc() . "locus-type", "locus type", "Specifies the type of locus described by the given entry") . parent::describeProperty($this->getVoc() . "locus-group", "locus group", "Groups locus types together into related sets. Below is a list of groups and the locus types within the group"));
}
if (!empty($previous_symbols)) {
$previous_symbols = explode(", ", $previous_symbols);
foreach ($previous_symbols as $previous_symbol) {
$previous_symbol_uri = "hgnc.symbol:" . $previous_symbol;
parent::AddRDF(parent::describeIndividual($previous_symbol_uri, $previous_symbol, parent::getVoc() . "Previous-Symbol") . parent::describeClass(parent::getVoc() . "Previous-Symbol", "Previous Symbol") . parent::triplify($id_res, $this->getVoc() . "previous-symbol", $previous_symbol_uri) . parent::describeProperty($this->getVoc() . "previous-symbol", "HGNC previous symbol", "Symbols previously approved by the HGNC for this gene"));
}
}
if (!empty($previous_names)) {
$previous_names = explode(", ", $previous_names);
foreach ($previous_names as $previous_name) {
$previous_name = str_replace("\"", "", $previous_name);
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "previous-name", utf8_encode(htmlspecialchars($previous_name))) . parent::describeProperty($this->getVoc() . "previous-name", "HGNC previous name", "Gene names previously approved by the HGNC for this gene"));
}
}
if (!empty($synonyms)) {
$synonyms = explode(", ", $synonyms);
foreach ($synonyms as $synonym) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "synonym", utf8_encode(htmlspecialchars($synonym))) . parent::describeProperty($this->getVoc() . "synonym", "synonym", "Other symbols used to refer to this gene"));
}
}
if (!empty($name_synonyms)) {
$name_synonyms = explode(", ", $name_synonyms);
foreach ($name_synonyms as $name_synonym) {
$name_synonym = str_replace("\"", "", $name_synonym);
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "name-synonym", utf8_encode(htmlspecialchars($name_synonym))) . parent::describeProperty($this->getVoc() . "name-synonym", "name synonym", "Other names used to refer to this gene"));
}
}
if (!empty($chromosome)) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "chromosome", utf8_encode(htmlspecialchars($chromosome))) . parent::describeProperty($this->getVoc() . "chromosome", "chromosome", "Indicates the location of the gene or region on the chromosome"));
}
if (!empty($date_approved)) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "date-approved", $date_approved, "xsd:date") . parent::describeProperty($this->getVoc() . "date-approved", "date approved", "Date the gene symbol and name were approved by the HGNC"));
}
if (!empty($date_modified)) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "date-modified", $date_modified, "xsd:date") . parent::describeProperty($this->getVoc() . "date-modified", "date modified", "the date the entry was modified by the HGNC"));
}
if (!empty($date_symbol_changed)) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "date-symbol-changed", $date_symbol_changed, "xsd:date") . parent::describeProperty($this->getVoc() . "date-symbol-changed", "date symbol changed", "The date the gene symbol was last changed by the HGNC from a previously approved symbol. Many genes receive approved symbols and names which are viewed as temporary (eg C2orf#) or are non-ideal when considered in the light of subsequent information. In the case of individual genes a change to the name (and subsequently the symbol) is only made if the original name is seriously misleading"));
}
if (!empty($date_name_changed)) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "date-name-changed", $date_name_changed, "xsd:date") . parent::describeProperty($this->getVoc() . "date-name-changed", "date name changed", "The date the gene name was last changed by the HGNC from a previously approved name"));
}
if (!empty($accession_numbers)) {
$accession_numbers = explode(", ", $accession_numbers);
foreach ($accession_numbers as $accession_number) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "accession", utf8_encode(htmlspecialchars($accession_number))) . parent::describeProperty($this->getVoc() . "accession", "accession number", "Accession numbers for each entry selected by the HGNC"));
}
}
if (!empty($enzyme_ids)) {
$enzyme_ids = explode(", ", $enzyme_ids);
foreach ($enzyme_ids as $enzyme_id) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "x-ec", utf8_encode(htmlspecialchars($enzyme_id))) . parent::describeProperty($this->getVoc() . "x-ec", "Enzyme Commission (EC) number", "Enzyme entries have Enzyme Commission (EC) numbers associated with them that indicate the hierarchical functional classes to which they belong"));
}
}
if (!empty($entrez_gene_id)) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-ncbigene", "ncbigene:{$entrez_gene_id}") . parent::describeProperty($this->getVoc() . "x-ncbigene", "NCBI Gene", "NCBI Gene provides curated sequence and descriptive information about genetic loci including official nomenclature, synonyms, sequence accessions, phenotypes, EC numbers, MIM numbers, UniGene clusters, homology, map locations, and related web sites"));
}
if (!empty($ensembl_gene_id)) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-ensembl", "ensembl:{$ensembl_gene_id}") . parent::describeProperty($this->getVoc() . "x-ensembl", "Ensembl Gene"));
}
if (!empty($mouse_genome_database_id)) {
if (strpos($mouse_genome_database_id, "MGI:") !== FALSE) {
$mouse_genome_database_id = substr($mouse_genome_database_id, 4);
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-mgi", "mgi:{$mouse_genome_database_id}") . parent::describeProperty($this->getVoc() . "x-mgi", "MGI entry"));
}
}
if (!empty($specialist_database_links)) {
$specialist_database_links = explode(", ", $specialist_database_links);
foreach ($specialist_database_links as $specialist_database_link) {
preg_match('/href="(\\S+)"/', $specialist_database_link, $matches);
if (!empty($matches[1])) {
parent::AddRDF(parent::QQuadO_URL($id_res, $this->getVoc() . "xref", $matches[1]) . parent::describeProperty($this->getVoc() . "xref", "Specialist database references."));
}
}
}
if (!empty($pubmed_ids)) {
$pubmed_ids = explode(", ", $pubmed_ids);
foreach ($pubmed_ids as $pubmed_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-pubmed", "pubmed:" . trim($pubmed_id)) . parent::describeProperty($this->getVoc() . "x-pubmed", "NCBI PubMed entry", "Identifier that links to published articles relevant to the entry in the NCBI's PubMed database."));
}
}
if (!empty($refseq_ids)) {
$refseq_ids = explode(", ", $refseq_ids);
foreach ($refseq_ids as $refseq_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-refseq", "refseq:" . trim($refseq_id)) . parent::describeProperty($this->getVoc() . "x-refseq", "NCBI Refseq entry", "The Reference Sequence (RefSeq) identifier for that entry, provided by the NCBI. As we do not aim to curate all variants of a gene only one selected RefSeq is displayed per gene report. RefSeq aims to provide a comprehensive, integrated, non-redundant set of sequences, including genomic DNA, transcript (RNA), and protein products. RefSeq identifiers are designed to provide a stable reference for gene identification and characterization, mutation analysis, expression studies, polymorphism discovery, and comparative analyses. In the HTML results page this ID links to the RefSeq page for that entry."));
}
}
if (!empty($gene_family_tag)) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "gene-family-tag", utf8_encode(htmlspecialchars($gene_family_tag))) . parent::describeProperty($this->getVoc() . "gene-family-tag", "Gene Family Tag", "Tag used to designate a gene family or group the gene has been assigned to, according to either sequence similarity or information from publications, specialist advisors for that family or other databases. Families/groups may be either structural or functional, therefore a gene may belong to more than one family/group. These tags are used to generate gene family or grouping specific pages at genenames.org and do not necessarily reflect an official nomenclature. Each gene family has an associated gene family tag and gene family description. If a particular gene is a member of more than one gene family, the tags and the descriptions will be shown in the same order."));
}
if (!empty($gene_family_description)) {
$gene_family_description = str_replace("\"", "", $gene_family_description);
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "gene-family-description", utf8_encode(htmlspecialchars($gene_family_description))) . parent::describeProperty($this->getVoc() . "gene-family-description", "gene family name", "Name given to a particular gene family. The gene family description has an associated gene family tag. Gene families are used to group genes according to either sequence similarity or information from publications, specialist advisors for that family or other databases. Families/groups may be either structural or functional, therefore a gene may belong to more than one family/group."));
}
if (!empty($record_type)) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "record-type", utf8_encode(htmlspecialchars($record_type))));
}
if (!empty($primary_ids)) {
$primary_ids = explode(", ", $primary_ids);
foreach ($primary_ids as $primary_id) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "primary-id", utf8_encode(htmlspecialchars($primary_id))) . parent::describeProperty($this->getVoc() . "primary-id", "primary identifier"));
}
}
if (!empty($secondary_ids)) {
$secondary_ids = explode(", ", $secondary_ids);
foreach ($secondary_ids as $secondary_id) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "secondary-id", utf8_encode(htmlspecialchars($secondary_id))) . parent::describeProperty($this->getVoc() . "secondary-id", "secondary identifier"));
}
}
if (!empty($ccd_ids)) {
$ccd_ids = explode(", ", $ccd_ids);
foreach ($ccd_ids as $ccd_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-ccds", "ccds:" . trim($ccd_id)) . parent::describeProperty($this->getVoc() . "x-ccds", "consensus CDS entry", "The Consensus CDS (CCDS) project is a collaborative effort to identify a core set of human and mouse protein coding regions that are consistently annotated and of high quality. The long term goal is to support convergence towards a standard set of gene annotations."));
}
}
if (!empty($vega_ids)) {
$vega_ids = explode(", ", $vega_ids);
foreach ($vega_ids as $vega_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-vega", "vega:" . trim($vega_id)) . parent::describeProperty($this->getVoc() . "x-vega", "VEGA gene entry"));
}
}
if (!empty($locus_specific_databases)) {
parent::AddRDF(parent::triplifyString($id_res, $this->getVoc() . "locus-specific-xref", utf8_encode(htmlspecialchars($locus_specific_databases))) . parent::describeProperty($this->getVoc() . "locus-specific-xref", "locus specific xref", "This contains a list of links to databases or database entries pertinent to the gene"));
}
if (!empty($entrez_gene_id_mappeddatasuppliedbyNCBI)) {
$entrez_gene_id_mappeddatasuppliedbyNCBI = explode(", ", $entrez_gene_id_mappeddatasuppliedbyNCBI);
foreach ($entrez_gene_id_mappeddatasuppliedbyNCBI as $gene_id) {
if (strstr($gene_id, ":") !== FALSE) {
$a = explode(":", $gene_id);
$gene_id = $a[1];
}
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-ncbigene", "ncbigene:" . trim($gene_id)) . parent::describeProperty($this->getVoc() . "x-ncbigene", "NCBI Gene entry"));
}
}
if (!empty($omim_id_mappeddatasuppliedbyNCBI)) {
$omim_id_mappeddatasuppliedbyNCBI = explode(", ", $omim_id_mappeddatasuppliedbyNCBI);
foreach ($omim_id_mappeddatasuppliedbyNCBI as $omim_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-omim", "omim:" . trim($omim_id)) . parent::describeProperty($this->getVoc() . "x-omim", "OMIM entry", "Identifier provided by Online Mendelian Inheritance in Man (OMIM) at the NCBI. This database is described as a catalog of human genes and genetic disorders containing textual information and links to MEDLINE and sequence records in the Entrez system, and links to additional related resources at NCBI and elsewhere. In the HTML results page this ID links to the OMIM page for that entry."));
}
}
if (!empty($refseq_mappeddatasuppliedbyNCBI)) {
$refseq_mappeddatasuppliedbyNCBI = explode(", ", $refseq_mappeddatasuppliedbyNCBI);
foreach ($refseq_mappeddatasuppliedbyNCBI as $refseq_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-refseq", "refseq:" . trim($refseq_id)) . parent::describeProperty($this->getVoc() . "x-refseq", "NCBI Refseq entry", "The Reference Sequence (RefSeq) identifier for that entry, provided by the NCBI. As we do not aim to curate all variants of a gene only one selected RefSeq is displayed per gene report. RefSeq aims to provide a comprehensive, integrated, non-redundant set of sequences, including genomic DNA, transcript (RNA), and protein products. RefSeq identifiers are designed to provide a stable reference for gene identification and characterization, mutation analysis, expression studies, polymorphism discovery, and comparative analyses. In the HTML results page this ID links to the RefSeq page for that entry."));
}
}
if (!empty($uniprot_id_mappeddatasuppliedbyUniProt)) {
$uniprot_id_mappeddatasuppliedbyUniProt = explode(", ", $uniprot_id_mappeddatasuppliedbyUniProt);
foreach ($uniprot_id_mappeddatasuppliedbyUniProt as $uniprot_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-uniprot", "uniprot:" . trim($uniprot_id)) . parent::describeProperty($this->getVoc() . "x-uniprot", "Uniprot entry", "The UniProt identifier, provided by the EBI. The UniProt Protein Knowledgebase is described as a curated protein sequence database that provides a high level of annotation, a minimal level of redundancy and high level of integration with other databases. In the HTML results page this ID links to the UniProt page for that entry."));
}
}
if (!empty($ensembl_id_mappeddatasuppliedbyEnsembl)) {
$ensembl_id_mappeddatasuppliedbyEnsembl = explode(", ", $ensembl_id_mappeddatasuppliedbyEnsembl);
foreach ($ensembl_id_mappeddatasuppliedbyEnsembl as $ensembl_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-ensembl", "ensembl:" . trim($refseq_id)) . parent::describeProperty($this->getVoc() . "x-ensembl", "Ensembl entry", "The Ensembl ID is derived from the current build of the Ensembl database and provided by the Ensembl team."));
}
}
if (!empty($ucsc_id_mappeddatasuppliedbyVega)) {
$ucsc_id_mappeddatasuppliedbyVega = explode(", ", $ucsc_id_mappeddatasuppliedbyVega);
foreach ($ucsc_id_mappeddatasuppliedbyVega as $vega_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-vega", "vega:" . trim($vega_id)) . parent::describeProperty($this->getVoc() . "x-vega", "Vega entry"));
}
}
if (!empty($ucsc_id_mappeddatasuppliedbyUCSC)) {
$ucsc_id_mappeddatasuppliedbyUCSC = explode(", ", $ucsc_id_mappeddatasuppliedbyUCSC);
foreach ($ucsc_id_mappeddatasuppliedbyUCSC as $ucsc_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-ucsc", "ucsc:" . trim($ucsc_id)) . parent::describeProperty($this->getVoc() . "x-ucsc", "UCSC entry"));
}
}
if (!empty($mouse_genome_database_id_mappeddatasuppliedbyMGI)) {
$mouse_genome_database_id_mappeddatasuppliedbyMGI = explode(", ", $mouse_genome_database_id_mappeddatasuppliedbyMGI);
foreach ($mouse_genome_database_id_mappeddatasuppliedbyMGI as $mgi_id) {
if (strpos($mgi_id, "MGI:") !== FALSE) {
$mgi_id = substr($mgi_id, 4);
}
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-mgi", "mgi:" . trim($mgi_id)) . parent::describeProperty($this->getVoc() . "x-mgi", "MGI entry"));
}
}
if (!empty($rat_genome_database_id_mappeddatasuppliedbyRGD)) {
$rat_genome_database_id_mappeddatasuppliedbyRGD = explode(", ", trim($rat_genome_database_id_mappeddatasuppliedbyRGD));
foreach ($rat_genome_database_id_mappeddatasuppliedbyRGD as $rgd_id) {
$rgd_id = trim($rgd_id);
if (!empty($rgd_id)) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-rgd", trim($rgd_id)) . parent::describeProperty($this->getVoc() . "x-rgd", "RGD entry"));
}
}
}
//write RDF to file
$this->WriteRDFBufferToWriteFile();
}
//while
}
function process()
{
$gb_record_str = "";
while ($aLine = $this->getReadFile()->Read(4096)) {
preg_match("/^\\/\\/\$/", $aLine, $matches);
if (count($matches)) {
//now remove the header if it is there
$gb_record_str = $this->removeHeader($gb_record_str);
$sectionsRaw = $this->parseGenbankRaw($gb_record_str);
/**
* SECTIONS being parsed:
* locus, definition, accession, version, keywords, segment, source, reference, features
*/
//get locus section(s)
$locus = $this->retrieveSections("LOCUS", $sectionsRaw);
$parsed_locus_arr = $this->parseLocus($locus);
//get the definition section
$definition = $this->retrieveSections("DEFINITION", $sectionsRaw);
$parsed_definition_arr = $this->parseDefinition($definition);
//get the accession
$accessions = $this->retrieveSections("ACCESSION", $sectionsRaw);
$parsed_accession_arr = $this->parseAccession($accessions);
//get the version
$versions = $this->retrieveSections("VERSION", $sectionsRaw);
$parsed_version_arr = $this->parseVersion($versions);
//get the keywords
$keywords = $this->retrieveSections("KEYWORDS", $sectionsRaw);
$parsed_keyword_arr = $this->parseKeywords($keywords);
//may not be any segment section
$segments = $this->retrieveSections("SEGMENT", $sectionsRaw);
if (!empty($segments)) {
$parsed_segments_arr = $this->parseSegment($segments);
}
$features = $this->retrieveSections("FEATURES", $sectionsRaw);
$parsed_features_arr = $this->parseFeatures($features);
//get the source section
$source = $this->retrieveSections("SOURCE", $sectionsRaw);
$parsed_source_arr = $this->parseSource($source);
$contig = $this->retrieveSections("CONTIG", $sectionsRaw);
if (!empty($contig)) {
$parsed_contig_arr = $this->parseContig($contig);
}
//get the reference section
$references = $this->retrieveSections("REFERENCE", $sectionsRaw);
$parsed_refs_arr = $this->parseReferences($references);
$gb_res = "gi:" . $parsed_version_arr['gi'];
$gb_label = utf8_encode(htmlspecialchars($parsed_definition_arr[0]));
parent::AddRDF(parent::describeIndividual($gb_res, $gb_label, $this->getVoc() . "genbank-record") . parent::triplifyString($gb_res, $this->getVoc() . 'sequence-length', $parsed_locus_arr[0]['sequence_length']) . parent::triplifyString($gb_res, $this->getVoc() . 'strandedness', $parsed_locus_arr[0]['strandedness']) . parent::triplify($gb_res, "rdf:type", $this->getRes() . $parsed_locus_arr[0]['mol_type']) . parent::triplifyString($gb_res, $this->getVoc() . 'chromosome-shape', $parsed_locus_arr[0]['chromosome_shape']) . parent::triplifyString($gb_res, $this->getVoc() . 'division-name', $parsed_locus_arr[0]['division_name']) . parent::triplifyString($gb_res, $this->getVoc() . 'date-of-entry', $parsed_locus_arr[0]['date']) . parent::triplifyString($gb_res, $this->getVoc() . 'source', utf8_encode($parsed_source_arr[0])) . parent::QQuadO_URL($gb_res, $this->getVoc() . 'fasta-seq', 'https://www.ncbi.nlm.nih.gov/sviewer/viewer.cgi?sendto=on&db=nucest&dopt=fasta&val=' . $parsed_version_arr['gi']));
foreach ($parsed_features_arr as $aFeature) {
//getFeatures
$type = $aFeature['type'];
$feat_desc = $this->getFeatures($type);
$label = preg_replace('/\\s\\s*/', ' ', $feat_desc['definition']);
$comment = null;
$value = $aFeature['value'];
$value_arr = explode("/", $value);
$location = preg_replace('/\\n/', '', $value_arr[0]);
$class_id = parent::getVoc() . md5($type);
$feat_res = parent::getRes() . md5($type . $location . $gb_res);
$feat_label = utf8_encode($type . " " . $location . " for " . $gb_res);
if (isset($feat_desc['comment'])) {
$comment = $feat_desc['comment'];
$comment = preg_replace('/\\s\\s*/', ' ', $comment);
$label .= " " . $comment;
}
parent::AddRDF(parent::describeClass($class_id, $label, parent::getVoc() . "Feature") . parent::describeIndividual($feat_res, $feat_label, $class_id) . parent::triplify($gb_res, $this->getVoc() . "has-feature", $feat_res));
foreach ($value_arr as $aL) {
//check if aL has an equals in it
$p = "/(\\S+)\\=(.*)/";
preg_match($p, $aL, $m);
if (count($m)) {
if ($m[1] == "db_xref") {
parent::AddRDF(parent::triplify($feat_res, "rdfs:seeAlso", str_replace("\"", "", $m[2])));
} else {
parent::AddRDF(parent::triplifyString($feat_res, $this->getVoc() . $m[1], utf8_encode(str_replace("\"", "", $m[2]))));
}
}
}
}
foreach ($parsed_accession_arr[0] as $acc) {
parent::AddRDF(parent::triplifyString($gb_res, $this->getVoc() . "accession", $acc));
}
if (isset($parsed_version_arr['versioned_accession'])) {
parent::AddRDF(parent::triplifyString($gb_res, $this->getVoc() . "versioned-accession", $parsed_version_arr['versioned_accession']));
}
if (isset($parsed_contig_arr)) {
foreach ($parsed_contig_arr as $aContig) {
parent::AddRDF(parent::triplifyString($gb_res, $this->getVoc() . "contig", parent::safeLiteral($aContig)));
}
}
foreach ($parsed_keyword_arr as $akw) {
parent::AddRDF(parent::triplifyString($gb_res, $this->getVoc() . "keyword", $akw));
}
if (isset($parsed_segments_arr)) {
foreach ($parsed_segments_arr as $aSeg) {
parent::AddRDF(parent::triplifyString($gb_res, $this->getVoc() . "segment-number", $aSeg['segment_number']) . parent::triplifyString($gb_res, $this->getVoc() . "total-segments", $aSeg['total_segments']));
}
}
foreach ($parsed_refs_arr as $aRef) {
$r = rand();
$ref_res = $this->getRes() . md5($r);
$ref_label = "reference for " . $gb_res;
if (isset($aRef['TITLE'])) {
parent::AddRDF(parent::describeIndividual($ref_res, $ref_label, $this->getVoc() . "reference") . parent::triplifyString($ref_res, $this->getVoc() . "title", $aRef['TITLE']));
}
if (isset($aRef['PUBMED'])) {
parent::AddRDF(parent::triplify($ref_res, $this->getVoc() . "x-pubmed", 'pubmed:' . $aRef['PUBMED']));
}
if (isset($aRef['AUTHORS'])) {
parent::AddRDF(parent::triplifyString($ref_res, $this->getVoc() . "authors", $aRef['AUTHORS']));
}
parent::AddRDF(parent::triplify($gb_res, $this->getVoc() . "reference", $ref_res) . parent::triplifyString($ref_res, $this->getVoc() . "coordinates", $aRef['COORDINATES']) . parent::triplifyString($ref_res, $this->getVoc() . "citation", $aRef['JOURNAL']));
}
$gb_record_str = "";
$this->WriteRDFBufferToWriteFile();
continue;
}
preg_match("/^\n\$/", $aLine, $matches);
if (count($matches) == 0) {
$gb_record_str .= $aLine;
}
}
//while
}
private function geneinfo()
{
$i = 1;
$header = $this->GetReadFile()->Read(200000);
while ($aLine = $this->GetReadFile()->Read(200000)) {
if ($i++ % 1000 == 0) {
parent::clear();
}
$a = $splitLine = explode("\t", $aLine);
if (count($splitLine) == 15) {
$taxid = "taxon:" . trim($splitLine[0]);
if (isset($this->taxids) and !isset($this->taxids[trim($splitLine[0])])) {
continue;
}
$aGeneId = trim($splitLine[1]);
$geneid = "ncbigene:" . trim($splitLine[1]);
$symbol = addslashes(stripslashes(trim($splitLine[2])));
$symbolid = "symbol:{$symbol}";
$locusTag = trim($splitLine[3]);
$symbols_arr = explode("|", $splitLine[4]);
$dbxrefs_arr = explode("|", $splitLine[5]);
$chromosome = trim($splitLine[6]);
$map_location = trim($splitLine[7]);
$description = addslashes(stripslashes(trim($splitLine[8])));
$type_of_gene = trim($splitLine[9]);
$symbol_authority = addslashes(stripslashes(trim($splitLine[10])));
$symbol_auth_full_name = addslashes(stripslashes(trim($splitLine[11])));
$nomenclature_status = addslashes(stripslashes(trim($splitLine[12])));
$other_designations = addslashes(stripslashes(trim($splitLine[13])));
$mod_date = date_parse(trim($splitLine[14]));
//check for a valid symbol
if ($symbol != "NEWENTRY") {
$this->AddRDF(parent::describeIndividual($geneid, "{$description} ({$symbolid}, {$taxid})", $this->getVoc() . "Gene") . parent::triplify($geneid, $this->getVoc() . "x-taxonomy", $taxid) . parent::triplifyString($geneid, $this->getVoc() . "symbol", $symbol) . parent::triplifyString($geneid, $this->getVoc() . "locus", addslashes(stripslashes($locusTag))) . parent::describeClass($this->getVoc() . "Gene", "NCBI Gene gene"));
if ($type_of_gene != '-') {
$this->AddRDF(parent::triplify($geneid, "rdf:type", $this->getVoc() . ucfirst($type_of_gene) . "-Gene") . parent::describeClass($this->getVoc() . ucfirst($type_of_gene) . "-Gene", ucfirst($type_of_gene) . " Gene"));
}
//symbol synonyms
foreach ($symbols_arr as $s) {
if ($s != "-") {
$this->AddRDF(parent::triplifyString($geneid, $this->getVoc() . "symbol-synonym", addslashes(stripslashes($s))));
}
}
//dbxrefs
foreach ($dbxrefs_arr as $dbx) {
if ($dbx != "-") {
$this->AddRDF(parent::triplifyString($geneid, $this->getVoc() . "dbxref", $dbx));
}
}
//chromosome
if ($chromosome != "-") {
$this->AddRDF(parent::triplifyString($geneid, $this->getVoc() . "chromosome", $chromosome));
}
//map location
if ($map_location != "-") {
$this->AddRDF(parent::triplifyString($geneid, $this->getVoc() . "map-location", $map_location));
}
//description
if ($description != "-") {
$this->AddRDF(parent::triplifyString($geneid, "dc:description", $description));
}
//nomenclature authority
if ($symbol_authority != "-") {
$this->AddRDF(parent::triplifyString($geneid, $this->getVoc() . "nomenclature-authority", $symbol_authority));
if ($symbol_auth_full_name != "-") {
$this->AddRDF(parent::triplifyString($geneid, $this->getVoc() . "nomenclature-authority-fullname", $symbol_auth_full_name));
}
}
//nomenclature status
if ($nomenclature_status != "-") {
$this->AddRDF(parent::triplifyString($geneid, $this->getVoc() . "nomenclature-status", $nomenclature_status));
}
//other designations
if ($other_designations != "-") {
foreach (explode("|", $other_designations) as $d) {
$this->AddRDF(parent::triplifyString($geneid, $this->getVoc() . "other-designation", $d));
}
}
//modification date
if ($mod_date != "-") {
$this->AddRDF(parent::triplifyString($geneid, $this->getVoc() . "modification-date", $mod_date["year"] . "-" . $mod_date["month"] . "-" . $mod_date["day"]));
}
}
}
parent::writeRDFBufferToWriteFile();
}
// while
}
function gene_expression()
{
$h = explode(",", parent::getReadFile()->read());
$expected_columns = 8;
if (($n = count($h)) != $expected_columns) {
trigger_error("Found {$n} columns in gene file - expecting {$expected_columns}!", E_USER_WARNING);
return false;
}
while ($l = parent::getReadFile()->read(200000)) {
$data = str_getcsv($l);
$mgi_symbol = $data[0];
$mgi_description = $data[1];
$geneid = $data[2];
$total_datasets = $data[3];
$total_ovexp = $data[4];
$total_underexp = $data[5];
$p_value = $data[6];
$expression = $data[7];
$id = parent::getRes() . md5($geneid . $total_datasets . $total_ovexp . $total_underexp . $p_value . $expression);
$evidence_id = parent::getRes() . md5($geneid . $total_datasets . $total_ovexp . $total_underexp . $p_value . $expression . "_evidence");
$label = "Dietary restriction induced " . $expression . "-expression of " . $mgi_symbol . " based on microarray results from " . $total_datasets . " datasets, with p-value " . $p_value;
$type_label = "Gene " . ucfirst($expression) . " Expression";
$type = parent::getVoc() . str_replace(" ", "-", $type_label);
parent::addRDF(parent::describeIndividual($id, $label, $type) . parent::describeClass($type, $type_label) . parent::triplify($id, parent::getVoc() . "gene", "ncbigene:" . $geneid) . parent::triplifyString("ncbigene:" . $geneid, parent::getVoc() . "mgi-gene-symbol", $mgi_symbol) . parent::triplifyString("ncbigene:" . $geneid, parent::getVoc() . "mgi-gene-description", $mgi_description) . parent::triplify($id, parent::getVoc() . "evidence", $evidence_id) . parent::triplifyString($id, parent::getVoc() . "perturbation-context", "dietary restriction") . parent::triplifyString($evidence_id, parent::getVoc() . "total-number-datasets", $total_datasets) . parent::triplifyString($evidence_id, parent::getVoc() . "total-number-datasets-overexpressed", $total_ovexp) . parent::triplifyString($evidence_id, parent::getVoc() . "total-number-datasets-underexpressed", $total_underexp) . parent::triplifyString($evidence_id, parent::getVoc() . "p-value", $p_value));
parent::writeRDFBufferToWriteFile();
}
//while
}
/**
* Convert pubchem substance XML record to RDF
**/
function parse_substance_record(&$xml)
{
$root = $xml->GetXMLRoot();
// pubchem identifier and version
$sid = array_shift($root->xpath('//PC-Substance_sid/PC-ID/PC-ID_id'));
$sid_version = array_shift($root->xpath('//PC-Substance_sid/PC-ID/PC-ID_version'));
$psid = $this->getPcsNs() . $sid;
parent::addRDF(parent::describeIndividual($psid, null, $this->getPcsVoc() . "Substance"));
parent::addRDF(parent::triplifyString($psid, $this->getPcsVoc() . "version", parent::safeLiteral($sid_version)));
// reference to pubchem compounds
$pc_compounds = $root->xpath('//PC-Substance_compound/PC-Compounds/PC-Compound');
foreach ($pc_compounds as $compound) {
$cid = array_shift($compound->xpath('./PC-Compound_id/PC-CompoundType/PC-CompoundType_id/PC-CompoundType_id_cid'));
$cid_type = array_shift($compound->xpath('./PC-Compound_id/PC-CompoundType/PC-CompoundType_type'));
if ($cid != "") {
$pcid = $this->getPccNs() . $cid;
parent::addRDF(parent::triplify($psid, $this->getPcsVoc() . "compound", $pcid));
}
}
// database cross references (xref)
// source identifier
$source_id = array_shift($root->xpath('//PC-Substance_source/PC-Source/PC-Source_db/PC-DBTracking/PC-DBTracking_source-id/Object-id/Object-id_str'));
parent::addRDF(parent::triplifyString($psid, $this->getPcsVoc() . "source-identifier", parent::safeLiteral($source_id)));
// synonyms
$synonyms = $root->xpath('//PC-Substance_synonyms/PC-Substance_synonyms_E');
foreach ($synonyms as $synonym) {
parent::addRDF(parent::triplifyString($psid, $this->getPcsVoc() . "synonym", parent::safeLiteral($synonym)));
}
//comment
$comments = $root->xpath('//PC-Substance_comment/PC-Substance_comment_E');
foreach ($comments as $comment) {
if ($comment !== "") {
parent::addRDF(parent::triplifyString($psid, "rdfs:comment", parent::safeLiteral($comment)));
}
}
}
function process($file)
{
$z = 1;
while ($l = parent::getReadFile()->read(100000)) {
if ($z % 100000 == 0) {
parent::clear();
}
if ($l[0] == "!") {
continue;
}
$fields = explode("\t", $l);
if (count($fields) != 17) {
trigger_error("Expected 17 columns, but found " . count($fields), E_USER_ERROR);
return false;
}
//get the Go id
$db = $fields[0];
$id = $fields[1];
$symbol = $fields[2];
$qualifier = $fields[3];
$goid = substr($fields[4], 3);
$refs = $this->getDbReferences($fields[5]);
$eco = $this->getEvidenceCodeLabelArr($fields[6]);
$aspect = $this->getAspect($fields[8]);
$label = $fields[9];
$synonyms = explode("|", $fields[10]);
$taxid = $fields[12];
$date = $this->parseDate($fields[13]);
$assignedBy = $fields[14];
//entity id
$eid = $this->getdbURI($db, $id);
if (!$eid) {
print_r($fields);
continue;
}
parent::addRDF(parent::describeIndividual($eid, $label, parent::getVoc() . "GO-Annotation") . parent::describeClass(parent::getVoc() . "GO-Annotation", "GO Annotation") . parent::triplifyString($eid, parent::getVoc() . "symbol", $symbol));
parent::addRDF(parent::triplify($eid, parent::getVoc() . "x-taxonomy", $taxid));
foreach ($synonyms as $s) {
if (!empty($s)) {
parent::addRDF(parent::triplifyString($eid, parent::getVoc() . "synonym", $s));
}
}
$rel = $aspect;
if ($qualifier == 'NOT') {
if ($aspect == 'process') {
$rel = 'not-in-process';
}
if ($aspect == 'function') {
$rel = 'not-has-function';
}
if ($aspect == 'component') {
$rel = 'not-in-component';
}
}
parent::addRDF(parent::describeObjectProperty(parent::getVoc() . $rel, str_replace("-", " ", $rel)) . parent::triplify($eid, parent::getVoc() . $rel, "go:" . $goid));
$type = key($eco);
$aid = parent::getRes() . $file . "_" . $z++;
parent::addRDF(parent::describeObjectProperty(parent::getVoc() . "go-annotation", "GO annotation") . parent::triplify($eid, parent::getVoc() . "go-annotation", $aid));
$cat = parent::getRes() . md5($aspect);
parent::addRDF(parent::describeIndividual($aid, "{$id}-go:{$goid} association", parent::getVoc() . "GO-Annotation") . parent::triplify($aid, parent::getVoc() . "target", $eid) . parent::triplify($aid, parent::getVoc() . "go-term", "go:" . $goid) . parent::triplify($aid, parent::getVoc() . "evidence", "eco:" . $eco[$type][1]) . parent::triplify($aid, parent::getVoc() . "go-category", $cat) . parent::describeClass($cat, $aspect) . parent::triplifyString($aid, parent::getVoc() . "assigned-by", $assignedBy));
if ($date != '') {
parent::addRDF(parent::triplifyString($aid, parent::getVoc() . "entry-date", $date . "T00:00:00Z", "xsd:dateTime"));
}
foreach ($refs as $ref) {
$b = explode(":", $ref);
if ($b[0] == 'PMID') {
parent::addRDF(parent::triplify($aid, parent::getVoc() . "article", "pubmed:" . $b[1]));
}
}
//write RDF to file
parent::writeRDFBufferToWriteFile();
}
}
function CTD_Genes()
{
$first = true;
while ($l = $this->GetReadFile()->Read()) {
if ($l[0] == '#') {
continue;
}
$a = explode("\t", $l);
// check number of columns
if ($first) {
if (($c = count(explode("\t", $l))) != 8) {
trigger_error("CTD_genes function expects 8 fields, found {$c}!" . PHP_EOL, E_USER_WARNING);
return FALSE;
}
$first = false;
}
$symbol = str_replace(array("\\/"), array('|'), $a[0]);
$label = str_replace("\\+/", '+', $a[1]);
$geneid = "ncbigene:" . $a[2];
$synonyms = $a[4];
$this->addRDF(parent::describeIndividual($geneid, $label, $this->getVoc() . "Gene") . parent::triplifyString($geneid, $this->getVoc() . "gene-symbol", $symbol) . parent::describeClass($this->getVoc() . "Gene", "CTD Gene"));
$ids = array(3 => array('rel' => "alternative-ncbigene-id", 'ns' => "ncbigene"), 4 => array('rel' => 'synonym'), 5 => array('rel' => 'x-biogrid', 'ns' => 'biogrid'), 6 => array('rel' => 'x-pharmgkb', 'ns' => 'pharmgkb'), 7 => array('rel' => 'x-uniprot', 'ns' => 'uniprot'));
foreach ($ids as $i => $v) {
if (!trim($a[$i])) {
continue;
}
$b = explode("|", $a[$i]);
foreach ($b as $c) {
if (isset($v['ns'])) {
parent::addRDF(parent::triplify($geneid, parent::getVoc() . $v['rel'], $v['ns'] . ":" . $c));
} else {
parent::addRDF(parent::triplifyString($geneid, parent::getVoc() . $v['rel'], $c));
}
}
}
parent::WriteRDFBufferToWriteFile();
}
return TRUE;
}
/**
* add an RDF representation of the incoming param to the model.
* @$qual_record_arr is an assoc array with the contents of one qualifier record
*/
private function makeQualifierRecordRDF($qual_record_arr)
{
//get the UI of the qualifier record
$qr_ui = $qual_record_arr["UI"][0];
$qr_res = $this->getNamespace() . $qr_ui;
$qr_label = $qual_record_arr['SH'][0];
parent::AddRDF(parent::describeIndividual($qr_res, $qr_label, $this->getVoc() . "Qualifier-Descriptor", $qr_label) . parent::describeClass($this->getVoc() . "Qualifier-Descriptor", "MeSH Qualifier Descriptor"));
//now get the descriptor_data_elements
$qde = $this->getQualifierDataElements();
//iterate over the properties
foreach ($qual_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 $anAn) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde["AN"], $anAn) . parent::describeProperty($this->getVoc() . $qde["AN"], "Relationship between a qualifier record and its annotation"));
}
//foreach
}
//foreach
}
//if
if ($k == "DA") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['DA'], $this->formatDate($vv), "xsd:date") . parent::describeProperty($this->getVoc() . $qde['DA'], "Relationship between a qualifier record and its date of entry"));
}
}
//if
if ($k == "DQ") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['DQ'], $this->formatDate($vv), "xsd:date") . parent::describeProperty($this->getVoc() . $qde['DQ'], "Relationship between a qualifier record and its date qualifier established"));
}
}
//if
if ($k == "GM") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['GM'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['GM'], "Relationship between a qualifier record and its grateful med note"));
}
}
if ($k == "HN") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['HN'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['HN'], "Relationship between a qualifier record and its history note"));
}
}
if ($k == "HN") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['HN'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['HN'], "Relationship between a qualifier record and its history note"));
}
}
if ($k == "MED") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['MED'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['MED'], "Relationship between a qualifier record and its backfile postings"));
}
}
if ($k == "M94") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['M94'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['M94'], "Relationship between a qualifier record and its backfile postings"));
}
}
if ($k == "M90") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['M90'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['M90'], "Relationship between a qualifier record and its backfile postings"));
}
}
if ($k == "M85") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['M85'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['M85'], "Relationship between a qualifier record and its backfile postings"));
}
}
if ($k == "M80") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['M80'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['M80'], "Relationship between a qualifier record and its backfile postings"));
}
}
if ($k == "M75") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['M75'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['M75'], "Relationship between a qualifier record and its backfile postings"));
}
}
if ($k == "M66") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['M66'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['M66'], "Relationship between a qualifier record and its backfile postings"));
}
}
if ($k == "MR") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['MR'], $this->formatDate($vv), "xsd:date") . parent::describeProperty($this->getVoc() . $qde['MR'], "Relationship between a qualifier record and its major revision date"));
}
}
//if
if ($k == "MS") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['MS'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['MS'], "Relationship between a qualifier record and its MeSH scope note"));
}
}
if ($k == "OL") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['OL'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['OL'], "Relationship between a qualifier record and its online note"));
}
}
if ($k == "QA") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['QA'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['QA'], "Relationship between a qualifier record and its toplical qualifier abbreviation"));
}
}
if ($k == "QE") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['QE'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['QE'], "Relationship between a qualifier record and its qualifier entry version"));
}
}
if ($k == "QS") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['QS'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['QS'], "Relationship between a qualifier record and its qualifier sort version"));
}
}
if ($k == "QT") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['QT'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['QT'], "Relationship between a qualifier record and its qualifier type"));
}
}
if ($k == "QX") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['QX'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['QX'], "Relationship between a qualifier record and its qualifier cross reference"));
}
}
if ($k == "RECTYPE") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['RECTYPE'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['RECTYPE'], "Relationship between a qualifier record and its record type"));
}
}
if ($k == "SH") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['SH'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['SH'], "Relationship between a qualifier record and its subheading"));
}
}
if ($k == "TN") {
foreach ($v as $kv => $vv) {
parent::AddRDF(parent::triplifyString($qr_res, $this->getVoc() . $qde['TN'], utf8_encode(htmlspecialchars($vv))) . parent::describeProperty($this->getVoc() . $qde['TN'], "Relationship between a qualifier record and its tree node allowed"));
}
}
} else {
trigger_error("Please add key to qualifier record map: " . $k . PHP_EOL, E_USER_ERROR);
}
//else
$this->WriteRDFBufferToWriteFile();
}
//foreach
$this->WriteRDFBufferToWriteFile();
}
function Parse($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 gene_xrefs()
{
while ($aLine = $this->getReadFile()->Read(4096)) {
$tLine = explode("\t", $aLine);
if (!$this->startsWith($tLine[0], "#")) {
$chromosome = null;
$cosmid = array();
$start_coord = null;
$gene_symbol = null;
$end_coord = null;
$strand = array();
$gene_location = array();
$ensembl_id = array();
$gene_id = null;
$ipi_ids = array();
$uniprotkb_ids = array();
$uniprot_tre = array();
$ensembl_peptide_id = array();
$refseq_ids = array();
$tair_ids = array();
$hinv_ids = array();
$unigene_ids = array();
$ccds_ids = array();
$refseq_gis = array();
$vega_genes = array();
$vega_peptides = array();
if (count(isset($tLine[0]))) {
@($chr_arr = readIdentifiers($tLine[0]));
if ($chr_arr[0] != "Un") {
$chromosome = $chr_arr[0];
}
}
if (count(isset($tLine[1]))) {
@($cosmid = readIdentifiers($tLine[1]));
}
if (count(isset($tLine[2]))) {
@($start_coord_t = readIdentifiers($tLine[2]));
if (count($start_coord_t) == 1) {
$start_coord = $start_coord_t[0];
}
}
if (count(isset($tLine[3]))) {
@($end_coord_t = readIdentifiers($tLine[3]));
if (count($end_coord_t) == 1) {
$end_coord = $end_coord_t[0];
}
}
if (count(isset($tLine[4]))) {
@($strand = readIdentifiers($tLine[4]));
}
if (count(isset($tLine[5]))) {
@($gene_location = readIdentifiers($tLine[5]));
}
if (count(isset($tLine[6]))) {
@($ensembl_id = readIdentifiers($tLine[6]));
}
if (count(isset($tLine[8]))) {
@($gene_id_t = readIdentifiers($tLine[8]));
if (count($gene_id_t) == 2) {
$gene_id = $gene_id_t[0];
$gene_symbol = $gene_id_t[1];
}
}
if (count(isset($tLine[9]))) {
@($ipi_ids = readIdentifiers($tLine[9]));
}
if (count(isset($tLine[10]))) {
@($uniprotkb_ids = readIdentifiers($tLine[10]));
}
if (count(isset($tLine[11]))) {
@($uniprot_tre = readIdentifiers($tLine[11]));
}
if (count(isset($tLine[12]))) {
@($ensembl_peptide_id = readIdentifiers($tLine[12]));
}
if (count(isset($tLine[13]))) {
@($refseq_ids = readIdentifiers($tLine[13]));
}
if (count(isset($tLine[14]))) {
@($tair_ids = readIdentifiers($tLine[14]));
}
if (count(isset($tLine[15]))) {
@($hinv_ids = readIdentifiers($tLine[15]));
}
if (count(isset($tLine[16]))) {
@($unigene_ids = readIdentifiers($tLine[16]));
}
if (count(isset($tLine[17]))) {
@($ccds_ids = readIdentifiers($tLine[17]));
}
if (count(isset($tLine[18]))) {
@($refseq_gis = readIdentifiers($tLine[18]));
}
if (count(isset($tLine[19]))) {
@($vega_genes = readIdentifiers($tLine[19]));
}
if (count(isset($tLine[20]))) {
@($refseq_ids = readIdentifiers($tLine[20]));
}
//lets make some RDF
if (count($gene_id)) {
$res = "gene:" . $gene_id;
parent::AddRDF(parent::triplifyString($res, $this->getVoc() . "gene-symbol", $gene_symbol) . parent::triplifyString($res, $this->getVoc() . "chromosome", $chromosome) . parent::triplifyString($res, $this->getVoc() . "start-coordinate", $start_coord) . parent::triplifyString($res, $this->getVoc() . "end-coordinate", $end_coord));
}
if (count($strand)) {
parent::AddRDF(parent::triplifyString($res, $this->getVoc() . "strand", $strand[0]));
}
if (count($ensembl_id)) {
foreach ($ensembl_id as $x) {
parent::AddRDF(parent::triplify($res, $this->getVoc() . "x-ensembl", "ensembl:" . $x));
}
}
if (count($gene_location)) {
foreach ($gene_location as $x) {
parent::AddRDF(parent::triplifyString($res, $this->getVoc() . "gene-location", $x));
}
}
if (count($ipi_ids)) {
foreach ($ipi_ids as $x) {
parent::AddRDF(parent::triplify($res, $this->getVoc() . "x-ipi", "ipi:" . $x));
}
}
if (count($uniprotkb_ids)) {
foreach ($uniprotkb_ids as $x) {
parent::AddRDF(parent::triplify($res, $this->getVoc() . "x-uniprot", "uniprot:" . $x));
}
}
if (count($uniprotkb_tre)) {
foreach ($uniprotkb_tre as $x) {
parent::AddRDF(parent::triplify($res, $this->getVoc() . "x-uniprot", "uniprot:" . $x));
}
}
if (count($ensembl_peptide_id)) {
foreach ($ensembl_peptide_id as $x) {
parent::AddRDF(parent::triplify($res, $this->getVoc() . "x-ensembl", "ensembl:" . $x));
}
}
if (count($refseq_ids)) {
foreach ($refseq_ids as $x) {
parent::AddRDF(parent::triplify($res, $this->getVoc() . "x-refseq", "refseq:" . $x));
}
}
if (count($tair_ids)) {
foreach ($tair_ids as $x) {
parent::AddRDF(parent::triplify($res, $this->getVoc() . "x-tair", "tair:" . $x));
}
}
if (count($hinv_ids)) {
foreach ($hinv_ids as $x) {
parent::AddRDF(parent::triplify($res, $this->getVoc() . "x-hinv", "hinv:" . $x));
}
}
if (count($unigene_ids)) {
foreach ($unigene_ids as $x) {
parent::AddRDF(parent::triplify($res, $this->getVoc() . "x-unigene", "unigene:" . $x));
}
}
if (count($refseq_gis)) {
foreach ($refseq_gis as $x) {
if (count($x) != 0 && $x != "\n" && $x != "") {
parent::AddRDF(parent::triplify($res, $this->getVoc() . "x-refseq", "refseq:" . $x));
}
}
}
$this->WriteRDFBufferToWriteFile();
}
}
}
