/** Generate an n-triple statement */
function QQuad($subject, $predicate, $object, $graph = null)
{
global $gns;
$s = explode(":", $subject);
$p = explode(":", $predicate);
$o = explode(":", $object);
if (!isset($gns[$s[0]])) {
trigger_error("Invalid subject qname " . $s[0]);
exit;
}
if (!isset($gns[$p[0]])) {
trigger_error("Invalid predicate qname " . $p[0]);
exit;
}
if (!isset($gns[$o[0]])) {
trigger_error("Invalid object qname " . $o[0]);
exit;
}
return Quad($gns[$s[0]] . $s[1], $gns[$p[0]] . $p[1], $gns[$o[0]] . $o[1]);
}
function write_type_relation_type_counts($fh, $counts)
{
global $options;
if ($counts !== null) {
foreach ($counts as $pred => $count) {
fwrite($fh, Quad("http://bio2rdf.org/dataset_resource:" . md5($options['url']), "http://bio2rdf.org/dataset_vocabulary:has_type_relation_type_count", "http://bio2rdf.org/dataset_resource:" . md5($options['url'] . $pred . "type_relation_type_count")));
fwrite($fh, Quad("http://bio2rdf.org/dataset_resource:" . md5($options['url'] . $pred . "type_relation_type_count"), "http://bio2rdf.org/dataset_vocabulary:has_predicate", $pred));
fwrite($fh, Quad("http://bio2rdf.org/dataset_resource:" . md5($options['url'] . $pred . "type_relation_type_count"), "http://bio2rdf.org/dataset_vocabulary:has_subject_type", $count["count"]["subject_type"]));
fwrite($fh, QuadLiteral("http://bio2rdf.org/dataset_resource:" . md5($options['url'] . $pred . "type_relation_type_count"), "http://bio2rdf.org/dataset_vocabulary:has_subject_count", $count["count"]["subject_count"]));
fwrite($fh, Quad("http://bio2rdf.org/dataset_resource:" . md5($options['url'] . $pred . "type_relation_type_count"), "http://bio2rdf.org/dataset_vocabulary:has_object_type", $count["count"]["object_type"]));
fwrite($fh, QuadLiteral("http://bio2rdf.org/dataset_resource:" . md5($options['url'] . $pred . "type_relation_type_count"), "http://bio2rdf.org/dataset_vocabulary:has_object_count", $count["count"]["object_count"]));
}
}
}
function OBO2TTL($indir, $outdir, $file)
{
global $gns, $gns_backup, $options;
$gns = $gns_backup;
$infile = $indir . $file;
$outfile = $outdir . $file . '.ttl';
$in = gzopen($infile, "r");
if (FALSE === $in) {
trigger_error("unable to open " . $infile);
exit;
}
$out = fopen($outfile, "w");
if (FALSE === $out) {
trigger_error("unable to open " . $outfile);
exit;
}
echo "Converting {$infile} to {$outfile}" . PHP_EOL;
if (FALSE !== ($pos = strrpos($infile, '\\'))) {
$file = substr($infile, $pos + 1);
} else {
if (FALSE !== ($pos = strrpos($infile, '/'))) {
$file = substr($infile, $pos + 1);
} else {
$file = $infile;
}
}
$file .= ".ttl";
$pos = strpos($file, ".");
$ontology = substr($file, 0, $pos);
$furi = "bio2rdf_resource:file/{$file}";
$ouri = "registry:{$ontology}";
$header = N3NSHeader();
$buf = QQuad($furi, "rdf:type", "sio:Document");
$buf .= QQuadL($furi, "rdfs:label", "Turtle converted OBO file for {$ontology} ontology (obtained through NCBO Bioportal) [bio2rdf_resource:{$file}]");
$buf .= QQuadL($furi, "dc:creator", "Michel Dumontier");
$buf .= QQuadL($furi, "sio:encodes", $ouri);
$buf .= QQuad($ouri, "rdf:type", "owl:Ontology");
$buf .= QQuadL($ouri, "rdfs:label", "{$ontology} ontology");
$buf .= QQuad($ouri, "sio:is-encoded-by", $furi);
$tid = '';
$first = true;
$is_a = false;
$is_deprecated = false;
$min = $buf;
while ($l = gzgets($in)) {
$lt = trim($l);
if (strlen($lt) == 0) {
continue;
}
if ($lt[0] == '!') {
continue;
}
if (strstr($l, "[Term]")) {
// top level node?
if ($first == true) {
// ignore the first case
$first = false;
} else {
if ($tid != '' && $is_a == false && $is_deprecated == false) {
$t = QQuad($tid, "rdfs:subClassOf", "bio2rdf_vocabulary:Entity");
$buf .= $t;
$min .= $t;
}
}
$is_a = false;
$is_deprecated = false;
unset($typedef);
$term = '';
$tid = '';
continue;
} else {
if (strstr($l, "[Typedef]")) {
$is_a = false;
$is_deprecated = false;
unset($term);
$tid = '';
$typedef = '';
continue;
}
}
// to fix error in obo generator
$lt = str_replace("synonym ", "synonym: ", $lt);
$lt = preg_replace("/\\{.*\\} !/", " !", $lt);
$a = explode(" !", $lt);
if (isset($a[1])) {
$exc = trim($a[1]);
}
$a = explode(": ", trim($a[0]), 2);
// let's go
if (isset($intersection_of)) {
if ($a[0] != "intersection_of") {
$intersection_of .= ")]." . PHP_EOL;
$buf .= $intersection_of;
if ($options['minimal+']['value'] == 'true') {
$min .= $intersection_of;
}
unset($intersection_of);
}
}
if (isset($typedef)) {
if ($a[0] == "id") {
$c = explode(":", $a[1]);
if (count($c) == 1) {
$ns = "obo";
$id = $c[0];
} else {
$ns = strtolower($c[0]);
$id = $c[1];
}
$id = str_replace(array("(", ")"), array("_", ""), $id);
$tid = $ns . ":" . $id;
$header .= AddToGlobalNS($ns);
$buf .= QQuadL($tid, "dc:identifier", $tid);
} else {
if ($a[0] == "name") {
$buf .= QQuadL($tid, "rdfs:label", addslashes(stripslashes($a[1])) . " [{$tid}]");
} else {
if ($a[0] == "is_a") {
if (FALSE !== ($pos = strpos($a[1], "!"))) {
$a[1] = substr($a[1], 0, $pos - 1);
}
$buf .= QQuad($tid, "rdfs:subPropertyOf", "obo:" . strtolower($a[1]));
} else {
if ($a[0] == "is_obsolete") {
$buf .= QQuad($tid, "rdf:type", "owl:DeprecatedClass");
$is_deprecated = true;
} else {
if ($a[0][0] == "!") {
$a[0] = substr($a[0], 1);
}
$buf .= QQuadL($tid, "obo:{$a['0']}", str_replace('"', '', stripslashes($a[1])));
}
}
}
}
} else {
if (isset($term)) {
if ($a[0] == "is_obsolete" && $a[1] == "true") {
$t = QQuad($tid, "rdf:type", "owl:DeprecatedClass");
$t .= QQuad($tid, "rdfs:subClassOf", "owl:DeprecatedClass");
$min .= $t;
$buf .= $t;
$is_deprecated = true;
} else {
if ($a[0] == "id") {
ParseQNAME($a[1], $ns, $id);
$header .= AddToGlobalNS($ns);
$tid = $ns . ":" . $id;
$buf .= QQuad($tid, "rdfs:isDefinedBy", $ouri);
$buf .= QQuadL($tid, "dc:identifier", $tid);
} else {
if ($a[0] == "name") {
$t = QQuadL($tid, "rdfs:label", str_replace(array("\"", "'"), array("", "\\\\'"), stripslashes($a[1])) . " [{$tid}]");
$min .= $t;
$buf .= $t;
} else {
if ($a[0] == "def") {
$t = str_replace(array("'", "\"", "\\", "\\\\'"), array("\\\\'", "", "", ""), $a[1]);
$min .= QQuadL($tid, "dc:description", $t);
$buf .= QQuadL($tid, "dc:description", $t);
} else {
if ($a[0] == "property_value") {
$b = explode(" ", $a[1]);
$buf .= QQuadL($tid, "obo_vocabulary:" . strtolower($b[0]), str_replace("\"", "", strtolower($b[1])));
} else {
if ($a[0] == "xref") {
// http://upload.wikimedia.org/wikipedia/commons/3/34/Anatomical_Directions_and_Axes.JPG
// Medical Dictionary:http\://www.medterms.com/
// KEGG COMPOUND:C02788 "KEGG COMPOUND"
// first get the comment
if (FALSE !== ($pos = strpos($a[1], '"'))) {
$comment = substr($a[1], $pos + 1, -1);
$identifier = substr($a[1], 0, $pos - 1);
} else {
$identifier = $a[1];
}
// next identify the namespace and identifier
if (FALSE !== ($pos = strpos($identifier, ":"))) {
$id = trim(substr($identifier, $pos + 1));
$raw_ns = strtolower(substr($identifier, 0, $pos));
// the raw ns is likely to be very dirty
// should map to the registry
// but for now, just add this namespace
$ns = str_replace(" ", "_", $raw_ns);
$header .= AddToGlobalNS($ns);
if (strstr($id, "http")) {
$buf .= Quad(GetFQURI($tid), GetFQURI("rdfs:seeAlso"), stripslashes($id));
} else {
$buf .= QQuad($tid, "rdfs:seeAlso", strtolower($ns) . ":" . str_replace(" ", " ", stripslashes($id)));
}
}
} else {
if ($a[0] == "synonym") {
// synonym: "entidades moleculares" RELATED [IUPAC:]
// synonym: "molecular entity" EXACT IUPAC_NAME [IUPAC:]
// synonym: "Chondrococcus macrosporus" RELATED synonym [NCBITaxonRef:Krzemieniewska_and_Krzemieniewski_1926]
//grab string inside double quotes
preg_match('/"(.*)"(.*)/', $a[1], $matches);
if (!empty($matches)) {
$a[1] = str_replace(array("\\", "\"", "'"), array("", "", "\\\\'"), $matches[1] . $matches[2]);
} else {
$a[1] = str_replace(array("\"", "'"), array("", "\\\\'"), $a[1]);
}
$rel = "SYNONYM";
$list = array("EXACT", "BROAD", "RELATED", "NARROW");
$found = false;
foreach ($list as $keyword) {
// get everything after the keyword up until the bracket [
if (FALSE !== ($k_pos = strpos($a[1], $keyword))) {
$str_len = strlen($a[1]);
$keyword_len = strlen($keyword);
$keyword_end_pos = $k_pos + $keyword_len;
$b1_pos = strrpos($a[1], "[");
$b2_pos = strrpos($a[1], "]");
$b_text = substr($a[1], $b1_pos + 1, $b2_pos - $b1_pos - 1);
$diff = $b1_pos - $keyword_end_pos - 1;
if ($diff != 0) {
// then there is more stuff here
$k = substr($a[1], $keyword_end_pos + 1, $diff);
$rel = trim($k);
} else {
// create the long predicate
$rel = $keyword . "_SYNONYM";
}
$found = true;
$str = substr($a[1], 0, $k_pos - 1);
break;
}
}
// check to see if we still haven't found anything
if ($found === false) {
// we didn't find one of the keywords
// so take from the start to the bracket
$b1_pos = strrpos($a[1], "[");
$str = substr($a[1], 0, $b1_pos - 1);
}
$rel = str_replace(" ", "_", $rel);
// $lit = addslashes($str.($b_text?" [".$b_text."]":""));
$l = QQuadL($tid, "obo_vocabulary:" . strtolower($rel), $str);
$buf .= $l;
} else {
if ($a[0] == "alt_id") {
ParseQNAME($a[1], $ns, $id);
if ($id != 'curators') {
$header .= AddToGlobalNS($ns);
$buf .= QQuad("{$ns}:{$id}", "rdfs:seeAlso", $tid);
}
} else {
if ($a[0] == "is_a") {
// do subclassing
ParseQNAME($a[1], $ns, $id);
$header .= AddToGlobalNS($ns);
$t = QQuad($tid, "rdfs:subClassOf", "{$ns}:{$id}");
$buf .= $t;
$min .= $t;
$is_a = true;
} else {
if ($a[0] == "intersection_of") {
if (!isset($intersection_of)) {
$intersection_of = GetFQURITTL($tid) . ' ' . GetFQURITTL('owl:equivalentClass') . ' [' . GetFQURITTL('rdf:type') . ' ' . GetFQURITTL('owl:Class') . '; ' . GetFQURITTL('owl:intersectionOf') . ' (';
}
/*
intersection_of: develops_from VAO:0000092 ! chondrogenic condensation
intersection_of: OBO_REL:has_part VAO:0000040 ! cartilage tissue
*/
$c = explode(" ", $a[1]);
if (count($c) == 1) {
// just a class
ParseQNAME($c[0], $ns, $id);
$header .= AddToGlobalNS($ns);
$intersection_of .= GetFQURITTL("{$ns}:{$id}");
} else {
if (count($c) == 2) {
// an expression
ParseQNAME($c[0], $pred_ns, $pred_id);
$header .= AddToGlobalNS($pred_ns);
ParseQNAME($c[1], $obj_ns, $obj_id);
$header .= AddToGlobalNS($obj_ns);
$intersection_of .= ' [' . GetFQURITTL('owl:onProperty') . ' ' . GetFQURITTL("obo:" . $pred_id) . '; ' . GetFQURITTL('owl:someValuesFrom') . ' ' . GetFQURITTL("{$obj_ns}:{$obj_id}") . '] ';
}
}
} else {
if ($a[0] == "relationship") {
if (!isset($relationship)) {
$relationship = GetFQURITTL($tid) . ' ' . GetFQURITTL('rdfs:subClassOf') . ' [' . GetFQURITTL('rdf:type') . ' ' . GetFQURITTL('owl:Class') . '; ' . GetFQURITTL('owl:intersectionOf') . ' (';
}
/*
relationship: develops_from VAO:0000092 ! chondrogenic condensation
relationship: OBO_REL:has_part VAO:0000040 ! cartilage tissue
*/
$c = explode(" ", $a[1]);
if (count($c) == 1) {
// just a class
ParseQNAME($c[0], $ns, $id);
$header .= AddToGlobalNS($ns);
$relationship .= GetFQURITTL("{$ns}:{$id}");
} else {
if (count($c) == 2) {
// an expression
ParseQNAME($c[0], $pred_ns, $pred_id);
$header .= AddToGlobalNS($pred_ns);
ParseQNAME($c[1], $obj_ns, $obj_id);
$header .= AddToGlobalNS($obj_ns);
$relationship .= ' [' . GetFQURITTL('owl:onProperty') . ' ' . GetFQURITTL("obo:" . $pred_id) . '; ' . GetFQURITTL('owl:someValuesFrom') . ' ' . GetFQURITTL("{$obj_ns}:{$obj_id}") . '] ';
$relationship .= ")]." . PHP_EOL;
$buf .= $relationship;
if ($options['minimal+']['value'] == 'true') {
$min .= $relationship;
}
unset($relationship);
}
}
} else {
// default handler
$buf .= QQuadL($tid, "obo:{$a['0']}", str_replace(array("\"", "'"), array("", "\\\\'"), stripslashes($a[1])));
}
}
}
}
}
}
}
}
}
}
}
} else {
//header
//format-version: 1.0
$a = explode(": ", trim($l));
$buf .= QQuadL($ouri, "obo:{$a['0']}", str_replace(array('"', '\\:'), array('\\"', ':'), isset($a[1]) ? $a[1] : ""));
}
}
fwrite($out, $header);
if ($options['minimal']['value'] == 'true' || $options['minimal+']['value'] == 'true') {
fwrite($out, $min);
} else {
fwrite($out, $buf);
}
$min = '';
$buf = '';
$header = '';
}
if (isset($intersection_of)) {
$buf .= $intersection_of . ")]." . PHP_EOL;
}
if (isset($relationship)) {
$buf .= $relationship . ")]." . PHP_EOL;
}
gzclose($in);
if ($options['minimal']['value'] == 'true' || $options['minimal+']['value'] == 'true') {
fwrite($out, $min);
} else {
fwrite($out, $buf);
}
fclose($out);
}
function addDatasetPropertyDatasetCount()
{
global $options;
$sparql = "SELECT DISTINCT ?p ?stype ?otype (COUNT(?s) AS ?n)\n" . $options['from-graph'] . " \n{\n\t?s ?p ?o .\n\t?s a ?stype .\n\t?o a ?otype .\n\tFILTER regex (?stype, \"vocabulary:Resource\")\n\tFILTER regex (?otype, \"vocabulary:Resource\")\n\tFILTER (?stype != ?otype)\n}";
$r = query($sparql);
foreach ($r as $c) {
$id = getID($c);
preg_match("/http:\\/\\/bio2rdf.org\\/([^_]+)_vocabulary/", $c->stype->value, $m1);
preg_match("/http:\\/\\/bio2rdf.org\\/([^_]+)_vocabulary/", $c->otype->value, $m2);
if (isset($m1[1]) and isset($m2[1])) {
$d1 = $m1[1];
$d2 = $m2[1];
$r = $c->p->value;
$label = "{$d1} connected to {$d2} through " . $c->n->value . " <{$r}> in " . $options['dataset_name'];
write(Quad($options['uri'], "http://rdfs.org/ns/void#subset", $id) . Quad($id, "http://www.w3.org/1999/02/22-rdf-syntax-ns#type", "http://rdfs.org/ns/void#LinkSet") . Quad($id, "http://www.w3.org/1999/02/22-rdf-syntax-ns#type", "http://bio2rdf.org/bio2rdf.dataset_vocabulary:Dataset-Dataset-Property-Dataset-Count") . QuadLiteral($id, "http://www.w3.org/2000/01/rdf-schema#label", $label, null, "en") . Quad($id, "http://rdfs.org/ns/void#linkPredicate", $c->p->value) . Quad($id, "http://rdfs.org/ns/void#subjectsTarget", $c->stype->value) . Quad($id, "http://rdfs.org/ns/void#objectsTarget", $c->otype->value) . QuadLiteral($id, "http://rdfs.org/ns/void#triples", $c->n->value, "long") . Quad("http://bio2rdf.org/bio2rdf.dataset_vocabulary:Dataset-Dataset-Property-Dataset-Count", "http://www.w3.org/2000/01/rdf-schema#subClassOf", "http://bio2rdf.org/bio2rdf.dataset_vocabulary:Dataset-Descriptor"));
}
}
}
function rdfize_counts($counts_arr, $dataset_name)
{
$rdf = "";
foreach ($counts_arr as $i => $nscount) {
$ns1 = $nscount['ns1'];
$ns2 = $nscount['ns2'];
$count = $nscount['count'];
$dataset_uri = "http://bio2rdf.org/dataset_resource:" . md5("http://" . $dataset_name . ".bio2rdf.org/sparql");
$count_uri = "http://bio2rdf.org/dataset_resource:" . md5($dataset_name . $ns1 . $ns2 . $count);
$rdf .= Quad($dataset_uri, "http://bio2rdf.org/dataset_vocabulary:has_nsns_count", $count_uri);
$rdf .= Quad($count_uri, "http://www.w3.org/1999/02/22-rdf-syntax-ns#type", "http://bio2rdf.org/dataset_vocabulary:Namespace_Namespace_Count");
$rdf .= Quad($count_uri, "http://bio2rdf.org/dataset_vocabulary:namespace1", "http://bio2rdf.org/" . $ns1);
$rdf .= Quad($count_uri, "http://bio2rdf.org/dataset_vocabulary:namespace2", "http://bio2rdf.org/" . $ns2);
$rdf .= QuadLiteral($count_uri, "http://bio2rdforg/dataset_vocabulary:has_nsns_count_value", $count);
}
return $rdf;
}