function Run()
{
$idir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$files = parent::getParameterValue('files');
if ($files == 'all') {
$list = explode('|', parent::getParameterList('files'));
array_shift($list);
} else {
$list = explode(',', parent::getParameterValue('files'));
}
$dataset_description = '';
foreach ($list as $item) {
$lfile = $idir . $item . '.rpt';
$rfile = parent::getParameterValue('download_url') . $item . '.rpt';
if (!file_exists($lfile) || parent::getParameterValue('download') == 'true') {
echo "downloading {$item}...";
$ret = Utils::DownloadSingle($rfile, $lfile);
if ($ret != true) {
continue;
}
}
parent::setReadFile($lfile, true);
echo "Processing {$item}...";
$ofile = $odir . $item . '.' . parent::getParameterValue('output_format');
$gz = strstr(parent::getParameterValue('output_format'), "gz") ? true : false;
parent::setWriteFile($ofile, $gz);
$this->{$item}();
parent::getWriteFile()->close();
parent::getReadFile()->close();
echo "Done" . PHP_EOL;
parent::clear();
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("MGI {$item}")->setRetrievedDate(date("Y-m-d\\TH:i:s", filemtime($lfile)))->setFormat("text")->setPublisher("http://www.informatics.jax.org")->setHomepage("http://www.informatics.jax.org")->setRights("use")->setLicense("http://www.informatics.jax.org/mgihome/other/copyright.shtml")->setDataset("http://identifiers.org/mgi/");
$prefix = parent::getPrefix();
$bVersion = parent::getParameterValue('bio2rdf_release');
$date = date("Y-m-d\\TH:i:s");
$output_file = (new DataResource($this))->setURI("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$ofile}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$item} in {$prefix}")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/mgi/mgi.php")->setCreateDate($date)->setHomepage("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$prefix}.html")->setPublisher("http://bio2rdf.org")->setRights("use-share-modify")->setRights("by-attribution")->setRights("restricted-by-source-license")->setLicense("http://creativecommons.org/licenses/by/3.0/")->setDataset(parent::getDatasetURI());
if ($gz) {
$output_file->setFormat("application/gzip");
}
if (strstr(parent::getParameterValue('output_format'), "nt")) {
$output_file->setFormat("application/n-triples");
} else {
$output_file->setFormat("application/n-quads");
}
$dataset_description .= $source_file->toRDF() . $output_file->toRDF();
}
//foreach
// generate the dataset release file
$this->setWriteFile($odir . parent::getBio2RDFReleaseFile());
$this->getWriteFile()->write($dataset_description);
$this->getWriteFile()->close();
echo "done!" . PHP_EOL;
}
function run()
{
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$dd = '';
$files = parent::getParameterValue('files');
if ($files == 'all') {
$files = explode('|', parent::getParameterList('files'));
array_shift($files);
} else {
$files = explode(',', parent::getParameterValue('files'));
}
foreach ($files as $file) {
echo "processing {$file} ...";
$lfile = $ldir . $this->filemap[$file];
$rfile = parent::getParameterValue('download_url') . $this->filemap[$file];
if (!file_exists($lfile) || parent::getParameterValue('download') == 'true') {
$ret = utils::downloadSingle($rfile, $lfile);
if ($ret === false) {
echo "unable to download {$file} ... skipping" . PHP_EOL;
continue;
}
}
parent::setReadFile($lfile, true);
$suffix = parent::getParameterValue('output_format');
$ofile = "orphanet-" . $file . '.' . $suffix;
$gz = strstr(parent::getParameterValue('output_format'), "gz") ? $gz = true : ($gz = false);
/* parent::setWriteFile($odir.$ofile, $gz);
$this->$file($lfile);
parent::getWriteFile()->close();
*/
parent::getReadFile()->close();
parent::clear();
echo "done!" . PHP_EOL;
// dataset description
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("Orphanet: {$file}")->setRetrievedDate(parent::getDate(filemtime($lfile)))->setFormat("application/xml")->setPublisher("http://www.orpha.net")->setHomepage("http://www.orpha.net/")->setRights("use")->setRights("sharing-modified-version-needs-permission")->setLicense("http://creativecommons.org/licenses/by-nd/3.0/")->setDataset("http://identifiers.org/orphanet/");
$prefix = parent::getPrefix();
$bVersion = parent::getParameterValue('bio2rdf_release');
$date = parent::getDate(filemtime($odir . $ofile));
$output_file = (new DataResource($this))->setURI("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$ofile}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix}")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/orphanet/orphanet.php")->setCreateDate($date)->setHomepage("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$prefix}.html")->setPublisher("http://bio2rdf.org")->setRights("use-share-modify")->setRights("by-attribution")->setRights("restricted-by-source-license")->setLicense("http://creativecommons.org/licenses/by/3.0/")->setDataset(parent::getDatasetURI());
$gz = strstr(parent::getParameterValue('output_format'), ".gz") === FALSE ? false : true;
if ($gz) {
$output_file->setFormat("application/gzip");
}
if (strstr(parent::getParameterValue('output_format'), "nt")) {
$output_file->setFormat("application/n-triples");
} else {
$output_file->setFormat("application/n-quads");
}
$dd .= $source_file->toRDF() . $output_file->toRDF();
}
//foreach
parent::writeToReleaseFile($dd);
}
function parse()
{
// convert into json
$lfile = parent::getReadFile()->getFileName();
$xml = simplexml_load_file($lfile);
$json = json_encode($xml);
$db = json_decode($json, TRUE);
// miriam metadata
// $attributes = $db['@attributes'];
foreach ($db['datatype'] as $item) {
$this->parseItem($item);
}
}
function gene_interactions()
{
while ($l = parent::getReadFile()->Read()) {
if ($l[0] == '#') {
continue;
}
$data = explode("\t", $l);
if (count($data) != 11) {
trigger_error("Found " . count($data) . " columns, expecting 11");
continue;
}
$interaction = $data[0];
$interaction_type = str_replace("_", "-", $data[1]);
$interaction_type_label = str_replace("_", " ", $data[1]);
$int_additional_info = $data[2];
$gene1 = $data[5];
$gene2 = $data[8];
$interaction_id = parent::getNamespace() . $interaction;
if ($interaction_type == "Genetic") {
$int_pred = parent::getVoc() . "genetically-interacts-with";
} elseif ($interaction_type == "Physical") {
$int_pred = parent::getVoc() . "physically-interacts-with";
} elseif ($interaction_type == "Predicted") {
$int_pred = parent::getVoc() . "predicted-to-interact-with";
} elseif ($interaction_type == "Regulatory") {
$int_pred = parent::getVoc() . "regulates";
}
//elseif
if ($int_additional_info == "No_interaction") {
$interaction_label = "No " . strtolower($interaction_type) . " interaction between " . $gene1 . " and " . $gene2;
parent::addRDF(parent::describeIndividual($interaction_id, $interaction_label, parent::getVoc() . $interaction_type . "-Non-Interaction") . parent::describeClass(parent::getVoc() . $interaction_type . "-Non-Interaction", $interaction_type_label . " non-interaction") . parent::triplify($interaction_id, parent::getVoc() . "involves", parent::getNamespace() . $gene1) . parent::triplify($interaction_id, parent::getVoc() . "involves", parent::getNamespace() . $gene2));
$npa_id = parent::getRes() . md5($interaction_id . "negative property assertion");
$npa_label = "Negative property assertion stating that " . $gene1 . " and " . $gene2 . " do not have a " . $interaction_type_label . " interaction";
parent::addRDF(parent::describeIndividual($npa_id, $npa_label, "owl:NegativeObjectPropertyAssertion") . parent::triplify($npa_id, "owl:sourceIndividual", parent::getNamespace() . $gene1) . parent::triplify($npa_id, "owl:targetIndividual", parent::getNamespace() . $gene2) . parent::triplify($npa_id, "owl:assertionProperty", $int_pred));
} elseif ($int_additional_info == "N/A" || $int_additional_info == "Genetic_interaction") {
$interaction_label = $interaction_type . " interaction between " . $gene1 . " and " . $gene2;
parent::addRDF(parent::describeIndividual($interaction_id, $interaction_label, parent::getVoc() . $interaction_type . "-Interaction") . parent::describeClass(parent::getVoc() . $interaction_type . "-Interaction", $interaction_type_label . " Interaction") . parent::triplify($interaction_id, parent::getVoc() . "involves", parent::getNamespace() . $gene1) . parent::triplify($interaction_id, parent::getVoc() . "involves", parent::getNamespace() . $gene2) . parent::triplify(parent::getNamespace() . $gene1, $int_pred, parent::getNamespace() . $gene2));
} else {
$interaction_label = ($int_additional_info != "" ? $int_additional_info . " " : "") . strtolower($interaction_type) . " interaction between " . $gene1 . " and " . $gene2;
$type = parent::getVoc() . ($int_additional_info != "" ? $int_additional_info . "-" : "") . $interaction_type . "-Interaction";
$type_label = ($int_additional_info != "" ? $int_additional_info . " " : "") . $interaction_type_label . " Interaction";
parent::addRDF(parent::describeIndividual($interaction_id, $interaction_label, $type) . parent::describeClass($type, $type_label, parent::getVoc() . $interaction_type . "-Interaction") . parent::describeClass(parent::getVoc() . $interaction_type . "-Interaction", $interaction_type . " Interation") . parent::triplify($interaction_id, parent::getVoc() . "involves", parent::getNamespace() . $gene1) . parent::triplify($interaction_id, parent::getVoc() . "involves", parent::getNamespace() . $gene2) . parent::triplify(parent::getNamespace() . $gene1, $int_pred, parent::getNamespace() . $gene2));
}
//else
parent::WriteRDFBufferToWriteFile();
}
//while
}
function process()
{
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$rdir = parent::getParameterValue('download_url');
//which files are to be converted?
$files = trim($this->GetParameterValue('files'));
if ($files == 'all') {
$files = $this->getPackageMap();
} else {
$sel_arr = explode(",", $files);
$pm = $this->getPackageMap();
$files = array();
foreach ($sel_arr as $a) {
if (array_key_exists($a, $pm)) {
$files[$a] = $pm[$a];
}
}
}
if ($this->getParameterValue('limit_organisms') == true) {
$this->taxids = array_flip(explode(",", $this->getParameterValue('organisms')));
}
//set dataset graph to be dataset URI
$graph_uri = parent::getGraphURI();
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
$dataset_description = '';
//now iterate over the files array
foreach ($files as $module => $rfilename) {
$file = $module . ".gz";
$lfile = $ldir . $file;
$rfile = $rdir . $rfilename;
// download
if (!file_exists($lfile) || parent::getParameterValue('download') == true) {
trigger_error("{$lfile} not found. Will attempt to download.", E_USER_NOTICE);
$myfile = $lfile;
if ($module == "gene2sts" || $module == "gene2unigene") {
$myfile = "compress.zlib://" . $lfile;
}
echo "downloading {$module} ...";
utils::DownloadSingle($rfile, $myfile);
echo "done" . PHP_EOL;
}
}
foreach ($files as $module => $rfilename) {
$file = $module . ".gz";
$lfile = $ldir . $file;
$rfile = $rdir . $rfilename;
$ofile = $module . "." . parent::getParameterValue('output_format');
$gz = false;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
echo "Processing {$module} ... ";
parent::setReadFile($lfile, true);
parent::setWriteFile($odir . $ofile, $gz);
$fnx = $module;
if ($module == 'gene2refseq') {
$fnx = 'gene2accession';
}
$this->{$fnx}();
parent::clear();
echo 'done!' . PHP_EOL;
parent::getReadFile()->close();
parent::getWriteFile()->close();
// generate the dataset release file
// dataset description
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("NCBI Gene ({$module})")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat("text/tab-separated-value")->setFormat("application/gzip")->setPublisher("http://www.ncbi.nlm.nih.gov")->setHomepage("http://www.ncbi.nlm.nih.gov/gene")->setRights("use-share-modify")->setLicense("http://www.ncbi.nlm.nih.gov/About/disclaimer.html")->setDataset("http://identifiers.org/ncbigene/");
$prefix = parent::getPrefix();
$bVersion = parent::getParameterValue('bio2rdf_release');
$date = date("Y-m-d\\TG:i:s\\Z");
$output_file = (new DataResource($this))->setURI("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$ofile}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/ncbigene/ncbigene.php")->setCreateDate($date)->setHomepage("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$prefix}.html")->setPublisher("http://bio2rdf.org")->setRights("use-share-modify")->setRights("by-attribution")->setRights("restricted-by-source-license")->setLicense("http://creativecommons.org/licenses/by/3.0/")->setDataset(parent::getDatasetURI());
if ($gz) {
$output_file->setFormat("application/gzip");
}
if (strstr(parent::getParameterValue('output_format'), "nt")) {
$output_file->setFormat("application/n-triples");
} else {
$output_file->setFormat("application/n-quads");
}
$dataset_description .= $source_file->toRDF() . $output_file->toRDF();
}
//foreach
//set graph URI back to default value
parent::setGraphURI($graph_uri);
//write dataset description to file
echo "Generating dataset description... ";
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
}
function genes()
{
$h = explode("\t", parent::getReadFile()->read());
$expected_columns = 14;
if (($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)) {
$a = explode("\t", $l);
$id = parent::getNamespace() . $a[0];
$label = $a[3];
$this->genes[$a[0]] = $a[3];
parent::addRDF(parent::describeIndividual($id, $label, parent::getVoc() . "Gene") . parent::describeClass(parent::getVoc() . "Gene", "PharmGKB Gene"));
// link data
parent::addRDF(parent::triplify($id, "rdfs:seeAlso", "http://pharmgkb.org/gene/" . $a[0]) . parent::triplify($id, "rdfs:seeAlso", "http://www4.wiwiss.fu-berlin.de/diseasome/resource/genes/" . $a[0]) . parent::triplify($id, "rdfs:seeAlso", "http://dbpedia.org/resource/" . $a[0]));
if ($a[1]) {
parent::addRDF(parent::triplify($id, parent::getVoc() . "x-ncbigene", "ncbigene:" . $a[1]));
}
if ($a[2]) {
parent::addRDF(parent::triplify($id, parent::getVoc() . "x-ensembl", "ensembl:" . $a[2]));
}
if ($a[3]) {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "name", $a[3]) . parent::describeProperty(parent::getVoc() . "name", "Relationship between a PharmGKB entity and its name"));
}
if ($a[4]) {
parent::addRDF(parent::triplify($id, parent::getVoc() . "symbol", "symbol:" . $a[4]) . parent::describeProperty(parent::getVoc() . "symbol", "Relationship between a PharmGKB gene and a gene symbol"));
}
if ($a[5]) {
$b = explode('","', substr($a[5], 1, -2));
foreach ($b as $alt_name) {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "alternative-name", parent::safeLiteral(trim(stripslashes($alt_name)))));
}
parent::addRDF(parent::describeProperty(parent::getVoc() . "alternative-name", "Relationship between a PharmGKB gene and an alternative name"));
}
if ($a[6]) {
// these are not hgnc symbols
$b = explode('","', substr($a[6], 1, -2));
foreach ($b as $alt_symbol) {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "alternate-symbol", trim($alt_symbol)));
}
parent::addRDF(parent::describeProperty($id, parent::getVoc() . "alternate-symbol", "Relationship between a PharmGKB gene and an alternate gene symbol"));
}
if ($a[7]) {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "is-vip", $a[7]) . parent::describeProperty(parent::getVoc() . "is-vip", "Relationship between a PharmGKB gene and its vip status"));
}
if ($a[8]) {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "has-variant-annotation", $a[8]) . parent::describeProperty(parent::getVoc() . "has-variant-annotation", "Relationship between a PharmGKB gene and whether it has a variant annotation"));
}
if ($a[9]) {
$b = explode(",", $a[9]);
foreach ($b as $xref) {
$xref = trim($xref);
if (!$xref) {
continue;
}
$url = false;
$x = $this->MapXrefs($xref, $url, $ns, $id2);
$ns = str_replace(' ', '', $ns);
if ($url == true) {
parent::addRDF(parent::QQuadO_URL($id, parent::getVoc() . "x-{$ns}", $x));
} else {
parent::addRDF(parent::triplify($id, parent::getVoc() . "x-{$ns}", $x));
}
}
}
if ($a[10]) {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "cpic-dosing-guideline", $a[10]));
}
if ($a[11]) {
parent::addRDF(parent::triplifyString($id, parent::getVoc() . "chromosome", $a[11]) . parent::describeProperty(parent::getVoc() . "chrosomome", "Relationship between a PharmGKB gene and its chromosomal position") . parent::triplifyString($id, parent::getVoc() . "chromosome-start", $a[12]) . parent::triplifyString($id, parent::getVoc() . "chromosome-end", $a[13]));
}
parent::WriteRDFBufferToWriteFile();
}
}
function Parse($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 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($longevity_influence)));
parent::WriteRDFBufferToWriteFile();
}
}
function process($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 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", $p_value));
parent::writeRDFBufferToWriteFile();
}
//while
}
function process($db)
{
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
while ($l = parent::getReadFile()->read()) {
list($nsid, $name) = explode("\t", $l);
list($ns, $id) = explode(":", $nsid);
if (isset($this->idlist) and !in_array($id, $this->idlist)) {
continue;
}
if (isset($this->org)) {
$id = $ns . "_" . $id;
}
$uri = $this->getNamespace() . $id;
parent::addRDF(parent::describeIndividual($uri, $name, parent::getVoc() . ucfirst($db)) . parent::describeClass(parent::getVoc() . ucfirst($db), "KEGG {$db}") . parent::triplifyString($uri, parent::getVoc() . "internal-id", $nsid));
// now get the entries for each
$lfile = $ldir . $id . ".txt";
$rfile = parent::getParameterValue("download_url") . "get/{$nsid}";
if (!file_exists($lfile) || parent::getParameterValue('download') == 'true') {
echo "downloading {$nsid} ";
$ret = utils::downloadSingle($rfile, $lfile);
if ($ret === false) {
echo "unable to download " . $nsid . " ... skipping" . PHP_EOL;
continue;
}
echo "done. ";
}
echo "parsing {$nsid} ... ";
$this->parseEntry($lfile);
parent::writeRDFBufferToWriteFile();
if ($db === "pathway") {
$ko = str_replace("map", "ko", $id);
$lfile = $ldir . $id . ".kgml";
$rfile = "http://www.kegg.jp/kegg-bin/download?entry=" . $ko . "&format=kgml";
if (!file_exists($lfile) || parent::getParameterValue('download') == 'true') {
echo "downloading KGML for {$nsid} ";
$ret = utils::downloadSingle($rfile, $lfile);
if ($ret === false) {
echo "unable to download " . $nsid . " ... skipping" . PHP_EOL;
continue;
}
echo "done. ";
}
$this->parseKGML($lfile);
parent::writeRDFBufferToWriteFile();
}
echo "done!" . PHP_EOL;
}
}
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();
}
}
public function Run()
{
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
// make sure we have the zip archive
//which files are to be converted?
$selectedPackage = trim(parent::getParameterValue('files'));
if ($selectedPackage == 'all') {
$files = $this->getPackageMap();
} else {
$sel_arr = explode(",", $selectedPackage);
$pm = $this->getPackageMap();
$files = array();
foreach ($sel_arr as $a) {
if (array_key_exists($a, $pm)) {
$files[$a] = $pm[$a];
}
}
}
$dataset_description = '';
foreach ($files as $key => $value) {
$lfile = $ldir . $value['filename'];
if (!file_exists($lfile) && parent::getParameterValue('download') == false) {
trigger_error($lfile . " not found. Will attempt to download.", E_USER_NOTICE);
$this->SetParameterValue('download', true);
}
//download all files [except mapping file]
if ($this->GetParameterValue('download') == true) {
$rfile = $value["file_url"];
echo "downloading " . var_dump($value["file_url"]) . " ... ";
utils::downloadSingle($rfile, $lfile);
}
if ($key == "taxdmp" || $key == "gi2taxid_protein" || $key == "gi2taxid_nucleotide") {
//get the name of the zip archive
$lfile = $value["filename"];
// make sure we have the zip archive
$zinfile = $ldir . $lfile;
$zin = new ZipArchive();
if ($zin->open($zinfile) === FALSE) {
trigger_error("Unable to open {$zinfile}");
exit;
}
//now iterate over the files in the ziparchive
$source_file = (new DataResource($this))->setURI($value['file_url'])->setTitle('NCBI Taxonomy - ' . $key)->setRetrievedDate(date("Y-m-d\\TH:i:sP", filemtime($ldir . $lfile)))->setFormat('text/tab-separated-value')->setFormat('application/zip')->setPublisher('http://www.ncbi.nlm.nih.gov')->setHomepage('http://www.ncbi.nlm.nih.gov/taxonomy')->setRights('use')->setRights('attribution')->setLicense('https://www.nlm.nih.gov/copyright.html')->setDataset(parent::getDatasetURI());
$prefix = parent::getPrefix();
$bVersion = parent::getParameterValue('bio2rdf_release');
$date = date("Y-m-d\\TH:i:sP");
$output_file = (new DataResource($this))->setURI("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} - {$key}")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/taxonomy/taxonomy.php")->setCreateDate($date)->setHomepage("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$prefix}.html")->setPublisher("http://bio2rdf.org")->setRights("use-share-modify")->setRights("restricted-by-source-license")->setLicense("http://creativecommons/licenses/by/3.0/")->setDataset(parent::getDatasetURI());
$dataset_description .= $output_file->toRDF() . $source_file->toRDF();
foreach ($value["contents"] as $k => $fn) {
if ($k == "names" || $k == "nodes" || $k == "citations" || $k == "gencode" || $k == "division" || $k == "gi_taxid_prot" || $k == "gi_taxid_nucl") {
//if($k !== 'citations') continue;
$fpin = $zin->getStream($fn);
if (!$fpin) {
trigger_error("Unable to get pointer to {$fn} in {$zinfile}");
exit("failed\n");
}
$gzoutfile = $odir . "taxonomy-{$k}" . "." . parent::getParameterValue('output_format');
//set the write file
$gz = strstr(parent::getParameterValue('output_format'), 'gz') ? true : false;
parent::setReadFile($ldir . $lfile);
parent::getReadFile()->SetFilePointer($fpin);
parent::setWriteFile($gzoutfile, $gz);
echo "processing {$fn}...\n";
$this->{$k}();
$this->GetWriteFile()->Close();
echo "done!" . PHP_EOL;
parent::clear();
}
//if $k
}
//foreach
}
//if key taxdmp
$this->setWriteFile($odir . $this->getBio2RDFReleaseFile());
$this->getWriteFile()->write($dataset_description);
$this->getWriteFile()->close();
}
}
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", $freq_upper)));
parent::setCheckpoint('record');
}
parent::setCheckpoint('file');
}
function OBO2RDF($abbv)
{
$abbv = strtolower($abbv);
if ($abbv == "doid") {
$abbv = "do";
}
$minimal = parent::getParameterValue('detail') == 'min' ? true : false;
$minimalp = parent::getParameterValue('detail') == 'min+' ? true : false;
$version = parent::getParameterValue("bio2rdf_release");
$tid = '';
$first = true;
$is_a = false;
$is_deprecated = false;
$min = $buf = '';
$ouri = "http://bio2rdf.org/lsr:" . $abbv;
$dataset_uri = $abbv . "_resource:bio2rdf.dataset.{$abbv}.R" . $version;
parent::setGraphURI($dataset_uri);
$buf = parent::triplify($ouri, "rdf:type", "owl:Ontology");
$graph_uri = '<' . parent::getRegistry()->getFQURI(parent::getGraphURI()) . '>';
$bid = 1;
while ($l = parent::getReadFile()->read()) {
$lt = trim($l);
if (strlen($lt) == 0) {
continue;
}
if ($lt[0] == '!') {
continue;
}
if (strstr($l, "[Term]")) {
// first node?
if ($first == true) {
// ignore the first case
$first = false;
} else {
if ($tid != '' && $is_a == false && $is_deprecated == false) {
$t = parent::triplify($tid, "rdfs:subClassOf", "obo_vocabulary:Entity");
$buf .= $t;
$min .= $t;
}
}
$is_a = false;
$is_deprecated = false;
unset($typedef);
$term = '';
$tid = '';
continue;
} else {
if (strstr($l, "[Typedef]")) {
$is_a = false;
$is_deprecated = false;
unset($term);
$tid = '';
$typedef = '';
continue;
}
}
//echo "LINE: $l".PHP_EOL;
// to fix error in obo generator
$lt = str_replace("synonym ", "synonym: ", $lt);
$lt = preg_replace("/\\{.*\\} !/", " !", $lt);
$a = explode(" !", $lt);
if (isset($a[1])) {
$exc = trim($a[1]);
}
$a = explode(": ", trim($a[0]), 2);
// let's go
if (isset($intersection_of)) {
if ($a[0] != "intersection_of") {
// $intersection_of .= ")].".PHP_EOL;
//$buf .= $intersection_of;
if ($minimalp) {
$min .= $intersection_of;
}
unset($intersection_of);
}
}
if (isset($relationship)) {
if ($a[0] != "relationship") {
// $relationship .= ")].".PHP_EOL;
//$buf .= $relationship;
if ($minimalp) {
$min .= $relationship;
}
unset($relationship);
}
}
if (isset($typedef)) {
if ($a[0] == "id") {
$c = explode(":", $a[1]);
if (count($c) == 1) {
$ns = "obo";
$id = $c[0];
} else {
$ns = strtolower($c[0]);
$id = $c[1];
}
$id = str_replace(array("(", ")"), array("_", ""), $id);
$tid = $ns . ":" . $id;
} else {
if ($a[0] == "name") {
$buf .= parent::describeClass($tid, addslashes(stripslashes($a[1])));
} else {
if ($a[0] == "is_a") {
if (FALSE !== ($pos = strpos($a[1], "!"))) {
$a[1] = substr($a[1], 0, $pos - 1);
}
$buf .= parent::triplify($tid, "rdfs:subPropertyOf", "obo_vocabulary:" . strtolower($a[1]));
} else {
if ($a[0] == "is_obsolete") {
$buf .= parent::triplify($tid, "rdf:type", "owl:DeprecatedClass");
$is_deprecated = true;
} else {
if ($a[0][0] == "!") {
$a[0] = substr($a[0], 1);
}
$buf .= parent::triplifyString($tid, "obo_vocabulary:{$a['0']}", str_replace('"', '', stripslashes($a[1])));
}
}
}
}
} else {
if (isset($term)) {
if ($a[0] == "is_obsolete" && $a[1] == "true") {
$t = parent::triplify($tid, "rdf:type", "owl:DeprecatedClass");
$t .= parent::triplify($tid, "rdfs:subClassOf", "owl:DeprecatedClass");
$min .= $t;
$buf .= $t;
$is_deprecated = true;
} else {
if ($a[0] == "id") {
parent::getRegistry()->parseQName($a[1], $ns, $id);
$tid = "{$ns}:{$id}";
// $buf .= parent::describeClass($tid,null,"owl:Class");
// $buf .= parent::triplify($tid,"rdfs:isDefinedBy",$ouri);
} else {
if ($a[0] == "name") {
// $t = parent::triplifyString($tid,"rdfs:label",str_replace(array("\"", "'"), array("","\\\'"), stripslashes($a[1]))." [$tid]");
$label = str_replace(array("\"", "'"), array("", "\\\\'"), stripslashes($a[1]));
$t = parent::describeIndividual($tid, $label, "owl:Class");
$t .= parent::triplify($tid, "rdfs:isDefinedBy", $ouri);
$min .= $t;
$buf .= $t;
} else {
if ($a[0] == "def") {
$t = str_replace(array("'", "\"", "\\", "\\\\'"), array("\\\\'", "", "", ""), $a[1]);
$min .= parent::triplifyString($tid, "dc:description", $t);
$buf .= parent::triplifyString($tid, "dc:description", $t);
} else {
if ($a[0] == "property_value") {
$b = explode(" ", $a[1]);
$buf .= parent::triplifyString($tid, "obo_vocabulary:" . strtolower($b[0]), str_replace("\"", "", strtolower($b[1])));
} else {
if ($a[0] == "xref") {
// http://upload.wikimedia.org/wikipedia/commons/3/34/Anatomical_Directions_and_Axes.JPG
// Medical Dictionary:http\://www.medterms.com/
// KEGG COMPOUND:C02788 "KEGG COMPOUND"
// id-validation-regexp:\"REACT_[0-9\]\{1\,4}\\.[0-9\]\{1\,3}|[0-9\]+\"
//$a[1] = 'id-validation-regexp:\"REACT_[0-9\]\{1\,4}\\.[0-9\]\{1\,3}|[0-9\]+\"';
if (substr($a[1], 0, 4) == "http") {
$buf .= parent::triplify($tid, "rdfs:seeAlso", str_replace(array(" ", '"wiki"', "\\"), array("+", "", ""), $a[1]));
} else {
$b = explode(":", $a[1], 2);
if (substr($b[1], 0, 4) == "http") {
$buf .= parent::triplify($tid, "rdfs:seeAlso", stripslashes($b[1]));
} else {
$ns = str_replace(array(" ", "\\"), "", strtolower($b[0]));
$id = trim($b[1]);
// there may be a comment to remove
if (FALSE !== ($pos = strrpos($id, ' "'))) {
$comment = substr($id, $pos + 1, -1);
$id = substr($id, 0, $pos);
}
$id = stripslashes($id);
// there may be a source statement to remove
$id = preg_replace("/{.*\\}/", "", $id);
if ($ns == "pmid") {
$ns = "pubmed";
$y = explode(" ", $id);
$id = $y[0];
}
if ($ns == "xx") {
continue;
}
if ($ns == "icd9cm") {
$y = explode(" ", $id);
$id = $y[0];
}
if ($ns == "xref; umls_cui") {
continue;
}
if ($ns == "submitter") {
$ns = "chebi.submitter";
}
if ($ns == "wikipedia" || $ns == "mesh") {
$id = str_replace(" ", "+", $id);
}
if ($ns == "id-validation-regexp") {
$buf .= parent::triplifyString($tid, "obo_vocabulary:{$ns}", addslashes($id));
} else {
$buf .= parent::triplify($tid, "obo_vocabulary:x-{$ns}", "{$ns}:" . str_replace(" ", "-", $id));
}
}
}
} else {
if ($a[0] == "synonym") {
// synonym: "entidades moleculares" RELATED [IUPAC:]
// synonym: "molecular entity" EXACT IUPAC_NAME [IUPAC:]
// synonym: "Chondrococcus macrosporus" RELATED synonym [NCBITaxonRef:Krzemieniewska_and_Krzemieniewski_1926]
//grab string inside double quotes
preg_match('/"(.*)"(.*)/', $a[1], $matches);
if (!empty($matches)) {
$a[1] = str_replace(array("\\", "\"", "'"), array("", "", "\\\\'"), $matches[1] . $matches[2]);
} else {
$a[1] = str_replace(array("\"", "'"), array("", "\\\\'"), $a[1]);
}
$rel = "SYNONYM";
$list = array("EXACT", "BROAD", "RELATED", "NARROW");
$found = false;
foreach ($list as $keyword) {
// get everything after the keyword up until the bracket [
if (FALSE !== ($k_pos = strpos($a[1], $keyword))) {
$str_len = strlen($a[1]);
$keyword_len = strlen($keyword);
$keyword_end_pos = $k_pos + $keyword_len;
$b1_pos = strrpos($a[1], "[");
$b2_pos = strrpos($a[1], "]");
$b_text = substr($a[1], $b1_pos + 1, $b2_pos - $b1_pos - 1);
$diff = $b1_pos - $keyword_end_pos - 1;
if ($diff != 0) {
// then there is more stuff here
$k = substr($a[1], $keyword_end_pos + 1, $diff);
$rel = trim($k);
} else {
// create the long predicate
$rel = $keyword . "_SYNONYM";
}
$found = true;
$str = substr($a[1], 0, $k_pos - 1);
break;
}
}
// check to see if we still haven't found anything
if ($found === false) {
// we didn't find one of the keywords
// so take from the start to the bracket
$b1_pos = strrpos($a[1], "[");
$str = substr($a[1], 0, $b1_pos - 1);
}
$rel = str_replace(" ", "_", $rel);
// $lit = addslashes($str.($b_text?" [".$b_text."]":""));
$l = parent::triplifyString($tid, "obo_vocabulary:" . strtolower($rel), $str);
$buf .= $l;
} else {
if ($a[0] == "alt_id") {
parent::getRegistry()->parseQname($a[1], $ns, $id);
if ($id != 'curators') {
$buf .= parent::triplify("{$ns}:{$id}", "rdfs:seeAlso", stripslashes($tid));
}
} else {
if ($a[0] == "is_a") {
// do subclassing
parent::getRegistry()->parseQName($a[1], $ns, $id);
$t = parent::triplify($tid, "rdfs:subClassOf", "{$ns}:{$id}");
$buf .= $t;
$min .= $t;
$is_a = true;
} else {
if ($a[0] == "intersection_of") {
if (!isset($intersection_of)) {
// $intersection_of = '<'.parent::getRegistry()->getFQURI($tid).'> <'.parent::getRegistry()->getFQURI('owl:equivalentClass').'> [<'.parent::getRegistry()->getFQURI('rdf:type').'> <'.parent::getRegistry()->getFQURI('owl:Class').'>; <'.parent::getRegistry()->getFQURI('owl:intersectionOf').'> (';
$intersection_of = '<' . parent::getRegistry()->getFQURI($tid) . '> <' . parent::getRegistry()->getFQURI('owl:equivalentClass') . '> _:b' . ++$bid . " {$graph_uri} ." . PHP_EOL;
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('rdf:type') . '> <' . parent::getRegistry()->getFQURI('owl:Class') . "> {$graph_uri} ." . PHP_EOL;
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:intersectionOf') . '> _:b' . ++$bid . " {$graph_uri} ." . PHP_EOL;
}
/*
intersection_of: ECO:0000206 ! BLAST evidence
intersection_of: develops_from VAO:0000092 ! chondrogenic condensation
intersection_of: OBO_REL:has_part VAO:0000040 ! cartilage tissue
*/
$c = explode(" ", $a[1]);
if (count($c) == 1) {
// just a class
parent::getRegistry()->parseQName($c[0], $ns, $id);
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('rdfs:subClassOf') . '> <' . parent::getRegistry()->getFQURI("{$ns}:{$id}") . "> {$graph_uri} ." . PHP_EOL;
$buf .= parent::triplify($tid, "rdfs:subClassOf", "{$ns}:{$id}");
} else {
if (count($c) == 2) {
// an expression
parent::getRegistry()->parseQName($c[0], $pred_ns, $pred_id);
parent::getRegistry()->parseQName($c[1], $obj_ns, $obj_id);
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:onProperty') . '> <' . parent::getRegistry()->getFQURI("obo_vocabulary:" . $pred_id) . "> {$graph_uri} ." . PHP_EOL;
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:someValuesFrom') . '> <' . parent::getRegistry()->getFQURI("{$obj_ns}:{$obj_id}") . "> {$graph_uri} ." . PHP_EOL;
$buf .= parent::triplify($tid, "obo_vocabulary:{$pred_id}", "{$obj_ns}:{$obj_id}");
}
}
} else {
if ($a[0] == "relationship") {
if (!isset($relationship)) {
$relationship = '<' . parent::getRegistry()->getFQURI($tid) . '> <' . parent::getRegistry()->getFQURI('rdfs:subClassOf') . '> _:b' . ++$bid . " {$graph_uri} ." . PHP_EOL;
$relationship .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('rdf:type') . '> <' . parent::getRegistry()->getFQURI('owl:Class') . "> {$graph_uri} ." . PHP_EOL;
$relationship .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:intersectionOf') . '> _:b' . ++$bid . " {$graph_uri} ." . PHP_EOL;
}
/*
relationship: develops_from VAO:0000092 ! chondrogenic condensation
relationship: OBO_REL:has_part VAO:0000040 ! cartilage tissue
*/
$c = explode(" ", $a[1]);
if (count($c) == 1) {
// just a class
parent::getRegistry()->parseQName($c[0], $ns, $id);
$relationship .= parent::getRegistry()->getFQURI("{$ns}:{$id}");
$buf .= parent::triplify($tid, "rdfs:subClassOf", "{$ns}:{$id}");
} else {
if (count($c) == 2) {
// an expression
parent::getRegistry()->parseQName($c[0], $pred_ns, $pred_id);
parent::getRegistry()->parseQName($c[1], $obj_ns, $obj_id);
$relationship .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:onProperty') . '> <' . parent::getRegistry()->getFQURI("obo_vocabulary:" . $pred_id) . "> {$graph_uri} ." . PHP_EOL;
$relationship .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:someValuesFrom') . '> <' . parent::getRegistry()->getFQURI("{$obj_ns}:{$obj_id}") . "> {$graph_uri} ." . PHP_EOL;
$buf .= parent::triplify($tid, "obo_vocabulary:{$pred_id}", "{$obj_ns}:{$obj_id}");
}
}
} else {
// default handler
if (isset($a[1])) {
$buf .= parent::triplifyString($tid, "obo_vocabulary:{$a['0']}", str_replace(array("\"", "'"), array("", "\\\\'"), stripslashes($a[1])));
}
}
}
}
}
}
}
}
}
}
}
}
} else {
//header
//format-version: 1.0
$buf .= parent::triplifyString($ouri, "obo_vocabulary:{$a['0']}", str_replace(array('"', '\\:'), array('\\"', ':'), isset($a[1]) ? $a[1] : ""));
}
}
if ($minimal || $minimalp) {
parent::getWriteFile()->write($min);
} else {
parent::getWriteFile()->write($buf);
}
$min = '';
$buf = '';
$header = '';
}
//if(isset($intersection_of)) $buf .= $intersection_of.")].".PHP_EOL;
//if(isset($relationship)) $buf .= $relationship.")].".PHP_EOL;
if ($minimal || $minimalp) {
parent::getWriteFile()->Write($min);
} else {
parent::getWriteFile()->write($buf);
}
}
