function Parse()
{
$l = $this->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 = $this->GetReadFile()->Read(100000)) {
$a = explode("\t", trim($l));
// 13 is the original identifier
$ids = explode("|", $a[13], 2);
$this->GetNS()->ParsePrefixedName($ids[0], $ns, $str);
$this->Parse4IDLabel($str, $id, $label);
$id = str_replace('"', '', $id);
$iid = $this->GetNS()->MapQName("{$ns}:{$id}");
$this->AddRDF($this->QQuad($iid, "void:inDataset", $this->GetDatasetURI()));
// get the type
if ($a[52] == "X") {
$label = "Pairwise interaction between {$a['0']} and {$a['1']}";
$type = "Pairwise-Interaction";
} else {
if ($a[52] == "C") {
$label = $a[53] . " component complex";
$type = "Multimeric-Complex";
} else {
if ($a[52] == "Y") {
$label = "homomeric complex composed of {$a['0']}";
$type = "Homopolymeric-Complex";
}
}
}
$this->AddRDF($this->QQuad($iid, "rdf:type", "irefindex_vocabulary:{$type}"));
// generate the label
// interaction type[52] by method[6]
if ($a[6] != '-') {
$qname = $this->ParseString($a[6], $ns, $id, $method);
if ($qname) {
$this->AddRDF($this->QQuad($iid, "irefindex_vocabulary:method", $qname));
}
}
$method_label = '';
if ($method != 'NA' && $method != '-1') {
$method_label = " identified by {$method} ";
}
$this->AddRDF($this->QQuadL($iid, "rdfs:label", "{$label}" . $method_label . " [{$iid}]"));
$this->AddRDF($this->QQuadO_URL($iid, "rdfs:seeAlso", "http://wodaklab.org/iRefWeb/interaction/show/" . $a[50]));
// set the interators
for ($i = 0; $i <= 1; $i++) {
$p = 'a';
if ($i == 1) {
$p = 'b';
}
$interactor = $this->ParseString($a[$i], $ns, $id, $label);
$this->AddRDF($this->QQuad($iid, "irefindex_vocabulary:interactor_{$p}", $interactor));
// biological role
$role = $a[16 + $i];
if ($role != '-') {
$qname = $this->ParseString($role, $ns, $id, $label);
if ($qname != "mi:0000") {
$this->AddRDF($this->QQuad($iid, "irefindex_vocabulary:interactor_{$p}" . "_biological_role", $qname));
}
}
// experimental role
$role = $a[18 + $i];
if ($role != '-') {
$qname = $this->ParseString($role, $ns, $id, $label);
if ($qname != "mi:0000") {
$this->AddRDF($this->QQuad($iid, "irefindex_vocabulary:interactor_{$p}" . "_experimental_role", $qname));
}
}
// interactor type
$type = $a[20 + $i];
if ($type != '-') {
$qname = $this->ParseString($type, $ns, $id, $label);
$this->AddRDF($this->QQuad($interactor, "rdf:type", $qname));
}
}
// add the alternatives through the taxon + seq redundant group
for ($i = 2; $i <= 3; $i++) {
$taxid = '';
$irogid = "irefindex_irogid:" . $a[42 + ($i - 2)];
if (!isset($defined[$irogid])) {
$defined[$irogid] = '';
$this->AddRDF($this->QQuadL($irogid, "rdfs:label", "[{$irogid}]"));
$this->AddRDF($this->QQuad($irogid, "rdf:type", "irefindex_vocabulary:Taxon-Sequence-Identical-Group"));
$tax = $a[9 + ($i - 2)];
if ($tax && $tax != '-' && $tax != '-1') {
$taxid = $this->ParseString($tax, $ns, $id, $label);
$this->AddRDF($this->QQuad($irogid, "irefindex_vocabulary:taxon", $taxid));
}
}
$list = explode("|", $a[3]);
foreach ($list as $item) {
$qname = $this->ParseString($item, $ns, $id, $label);
if ($ns && $ns != 'irefindex_rogid' && $ns != 'irefindex_irogid') {
$this->AddRDF($this->QQuad($qname, "irefindex_vocabulary:taxon-sequence-identical-group", $irogid));
if ($taxid && $taxid != '-' && $taxid != '-1') {
$this->AddRDF($this->QQuad($qname, "irefindex_vocabulary:taxon", $taxid));
}
}
}
}
// add the aliases through the canonical group
for ($i = 4; $i <= 5; $i++) {
$icrogid = "irefindex_icrogid:" . $a[49 + ($i - 4)];
if (!isset($defined[$icrogid])) {
$defined[$icrogid] = '';
$this->AddRDF($this->QQuadL($icrogid, "rdfs:label", "[{$icrogid}]"));
$this->AddRDF($this->QQuad($icrogid, "rdf:type", "irefindex_vocabulary:Taxon-Sequence-Similar-Group"));
}
$list = explode("|", $a[3]);
foreach ($list as $item) {
$qname = $this->ParseString($item, $ns, $id, $label);
if ($ns && $ns != 'crogid' && $ns != 'icrogid') {
$this->AddRDF($this->QQuad($qname, "irefindex_vocabulary:taxon-sequence-similar-group", $icrogid));
}
}
}
// publications
$list = explode("|", $a[8]);
foreach ($list as $item) {
if ($item == '-') {
continue;
}
$qname = $this->ParseString($item, $ns, $id, $label);
$this->AddRDF($this->QQuad($iid, "irefindex_vocabulary:article", $qname));
}
// MI interaction type
if ($a[11] != '-' && $a[11] != 'NA') {
$qname = $this->ParseString($a[11], $ns, $id, $label);
$this->AddRDF($this->QQuad($iid, "rdf:type", $qname));
if (!isset($defined[$qname])) {
$defined[$qname] = '';
$this->AddRDF($this->QQuadL($qname, "rdfs:label", "{$label} [{$qname}]"));
}
}
// source
if ($a[12] != '-') {
$qname = $this->ParseString($a[12], $ns, $id, $label);
$this->AddRDF($this->QQuad($iid, "irefindex_vocabulary:source", $qname));
}
// confidence
$list = explode("|", $a[14]);
foreach ($list as $item) {
$this->ParseString($item, $ns, $id, $label);
if ($ns == 'lpr') {
// lowest number of distinct interactions that any one article reported
$this->AddRDF($this->QQuadL($iid, "irefindex_vocabulary:minimum-number-interactions-reported", $id));
} else {
if ($ns == "hpr") {
// higher number of distinct interactions that any one article reports
$this->AddRDF($this->QQuadL($iid, "irefindex_vocabulary:maximum-number-interactions-reported", $id));
} else {
if ($ns = 'hp') {
// total number of unique PMIDs used to support the interaction
$this->AddRDF($this->QQuadL($iid, "irefindex_vocabulary:number-supporting-articles", $id));
}
}
}
}
// expansion method
if ($a[15]) {
$this->AddRDF($this->QQuadL($iid, "irefindex_vocabulary:expansion-method", $a[15]));
}
// host organism
if ($a[28] != '-') {
$qname = $this->ParseString($a[28], $ns, $id, $label);
$this->AddRDF($this->QQuad($iid, "irefindex_vocabulary:host-organism", $qname));
}
// created
$this->AddRDF($this->QQuadL($iid, "dc:created", $a[30]));
// taxon-sequence identical interaction group
$this->AddRDF($this->QQuad($iid, "irefindex_vocabulary:taxon-sequence-identical-interaction-group", "irefindex_irigid:" . $a[44]));
// taxon-sequence similar interaction group
$this->AddRDF($this->QQuad($iid, "irefindex_vocabulary:taxon-sequence-similar-interaction-group", "irefindex_crigid:" . $a[50]));
$this->WriteRDFBufferToWriteFile();
}
}
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();
}
}
