public function Run()
{
$dataset_description = '';
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
//first get the files that are to be processed
$selectedPackage = trim(parent::getParameterValue('files'));
if ($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];
}
}
}
//download
if ($this->getParameterValue('download')) {
foreach ($files as $aP => $fn) {
foreach ($fn as $aFn) {
echo "downloading file {$aFn} :" . parent::getParameterValue('download_url') . $aFn . "..." . PHP_EOL;
file_put_contents($ldir . $aFn, file_get_contents(parent::getParameterValue('download_url') . $aFn));
}
}
}
//iterate over the files
$paths = $this->getFilePaths($ldir, 'gz');
$lfile = null;
foreach ($files as $k => $val) {
foreach ($val as $fn) {
if (in_array($fn, $paths)) {
$lfile = $fn;
$ofile = $odir . basename($fn, ".gz") . "." . parent::getParameterValue('output_format');
$gz = false;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
parent::setWriteFile($ofile, $gz);
parent::setReadFile($ldir . $lfile, true);
$source_file = (new DataResource($this))->setURI(parent::getParameterValue('download_url') . basename($fn))->setTitle('International Protein Index filename: ' . basename($fn))->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($ldir . $lfile)))->setFormat('text/ipi-format')->setFormat('application/zip')->setPublisher('https://www.ebi.ac.uk')->setHomepage('https://www.ebi.ac.uk/IPI')->setRights('use')->setRights('attribution')->setLicense('https://www.ebi.ac.uk')->setDataset(parent::getDatasetURI());
$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}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/ipi/ipi.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() . $source_file->toRDF();
echo "processing {$fn} ...";
$this->{$k}();
echo "done!" . PHP_EOL;
$this->setWriteFile($odir . $this->getBio2RDFReleaseFile());
$this->getWriteFile()->write($dataset_description);
$this->getWriteFile()->close();
}
}
}
}
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 Run()
{
$file = "homologene.data";
$ldir = $this->GetParameterValue('indir');
$odir = $this->GetParameterValue('outdir');
$rdir = $this->GetParameterValue('download_url');
$lfile = $ldir . $file;
if (!file_exists($lfile)) {
trigger_error($file . " not found. Will attempt to download.", E_USER_NOTICE);
parent::setParameterValue('download', true);
}
//download
$rfile = $rdir . $file;
if ($this->GetParameterValue('download') == true) {
echo "downloading {$file} ... ";
utils::downloadSingle($rfile, $lfile);
}
$ofile = 'homologene.' . parent::getParameterValue('output_format');
$gz = strstr(parent::getParameterValue('output_format'), "gz") ? $gz = true : ($gz = false);
parent::setReadFile($lfile);
parent::setWriteFile($odir . $ofile, $gz);
echo "processing {$file}... ";
$this->process();
echo "done!" . PHP_EOL;
parent::getWriteFile()->close();
// generate the dataset release file
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("NCBI Homologene")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat("text/tab-separated-value")->setPublisher("http://www.ncbi.nlm.nih.gov")->setHomepage("http://www.ncbi.nlm.nih.gov/homologene")->setRights("use-share-modify")->setLicense("http://www.ncbi.nlm.nih.gov/About/disclaimer.html")->setDataset("http://identifiers.org/homologene/");
$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 {$prefix}")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/homologene/homologene.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() . $output_file->toRDF();
echo "Generating dataset description... ";
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
}
function run()
{
$dataset_description = '';
$ldir = parent::GetParameterValue('indir');
$odir = parent::GetParameterValue('outdir');
//download
if ($this->GetParameterValue('download') == true) {
$list = $this->getFtpFileList('ftp.ncbi.nih.gov');
$total = count($list);
$counter = 1;
foreach ($list as $f) {
echo "downloading file {$counter} out of {$total} :" . parent::getParameterValue('download_url') . $f . "... " . PHP_EOL;
file_put_contents($ldir . $f, file_get_contents(parent::GetParameterValue('download_url') . $f));
$counter++;
}
}
//if download
//iterate over the files
$paths = $this->getFilePaths($ldir, 'gz');
$lfile = null;
foreach ($paths as $aPath) {
$lfile = $aPath;
$ofile = $odir . basename($aPath, ".gz") . "." . parent::getParameterValue('output_format');
$gz = false;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
parent::setWriteFile($ofile, $gz);
parent::setReadFile($ldir . $lfile, true);
$source_file = (new DataResource($this))->setURI(parent::getParameterValue('download_url') . basename($aPath))->setTitle('NCBI UniSTS filename: ' . basename($aPath))->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($ldir . $lfile)))->setFormat('xml/unists-format')->setFormat('application/zip')->setPublisher('https://www.ncbi.nlm.nih.gov')->setHomepage('https://www.ncbi.nlm.nih.gov/unists')->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\\TG:i:s\\Z");
$output_file = (new DataResource($this))->setURI("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/unists/unists.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() . $source_file->toRDF();
echo "processing {$aPath} ...";
$this->process();
echo "done!" . PHP_EOL;
$this->setWriteFile($odir . $this->getBio2RDFReleaseFile());
$this->getWriteFile()->write($dataset_description);
$this->getWriteFile()->close();
}
//foreach
}
function Run()
{
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
//check dependencies
$d = $this->checkDependencies();
if ($d) {
//build pdb2rdf
echo "building pdb2rdf..." . PHP_EOL;
$cmd = "mvn clean install -DskipTests -f " . __DIR__ . "/pom.xml";
$build_out = shell_exec($cmd);
$out_ver = $this->verifyMavenBuildOutput($build_out);
if ($out_ver) {
//now check if download is desired
if ($this->getParameterValue('download')) {
if (!$this->downloadFiles($ldir)) {
trigger_error("Not all files downloaded!", E_USER_WARNING);
}
}
//extract pdb2rdf-cli from the target directory
if (!$this->extractCli()) {
trigger_error("Could not extract pdb2rdf!", E_USER_ERROR);
}
//now get ready to run pdb2rdf.sh
if (!$this->runPdb2Rdf($ldir, $odir)) {
trigger_error("Could not run Pdb2RDF correctly!", E_USER_ERROR);
exit;
} else {
echo "done!\n";
}
} else {
trigger_error("Could not build pdb2rdf. Please try manually!", E_USER_ERROR);
}
} else {
trigger_error("Dependencies not met!", E_USER_ERROR);
exit;
}
}
function Run()
{
echo "processing miriam database";
// directory shortcuts
$ldir = $this->getParameterValue('indir');
$odir = $this->getParameterValue('outdir');
// download and set the read file
$file = 'miriam.xml';
$rfile = $this->getParameterValue("download_url");
$lfile = $ldir . $file;
if (!file_exists($lfile) || $this->getParameterValue("download") == "true") {
utils::downloadSingle($rfile, $lfile);
}
parent::setReadFile($lfile);
// set the write file
$outfile = "miriam." . parent::getParameterValue('output_format');
$gz = strstr(parent::getParameterValue('output_format'), ".gz") === FALSE ? false : true;
parent::setWriteFile(parent::getParameterValue("outdir") . $outfile, $gz);
$this->parse();
parent::WriteRDFBufferToWriteFile();
$this->getWriteFile()->Close();
return true;
}
function Run()
{
$idir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$files = parent::getParameterValue('files');
// set the work
if ($files != 'all') {
// check if comma-separated, or hyphen-range
$list = explode(",", $files);
if (count($list) == 1) {
// try hyphen separated
$range = explode("-", $files);
if (count($range) == 2) {
for ($i = $range[0]; $i <= $range[1]; $i++) {
$myfiles[] = $i;
}
} else {
// must a single entry
$myfiles[] = $files;
}
} else {
$myfiles = $list;
}
}
$rest_uri = 'http://sabiork.h-its.org/sabioRestWebServices/';
$getReactionIds_url = $rest_uri . "suggestions/SABIOReactionIDs";
$reaction_list_file = $idir . "reactions.xml";
if (!file_exists($reaction_list_file) || parent::getParameterValue('download') == 'true') {
$xml = file_get_contents($getReactionIds_url);
if (FALSE === $reaction_list_file) {
exit;
}
$f = new FileFactory($reaction_list_file);
$f->Write($xml);
$f->Close();
}
$xml = simplexml_load_file($reaction_list_file);
$total = count($xml->SABIOReactionID);
if (isset($myfiles)) {
$total = count($myfiles);
}
$i = 0;
parent::setCheckpoint('dataset');
$graph_uri = parent::getGraphURI();
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
$suffix = parent::getParameterValue('output_format');
$ofile = "sabiork." . $suffix;
$gz = strstr(parent::getParameterValue('output_format'), "gz") ? true : false;
parent::setWriteFile($odir . $ofile, $gz);
foreach ($xml->SABIOReactionID as $rid) {
parent::setCheckpoint('file');
if (isset($myfiles)) {
if (!in_array($rid, $myfiles)) {
continue;
}
}
$i++;
echo "{$i} / {$total} : reaction {$rid}" . PHP_EOL;
$reaction_file = $idir . "reaction_" . $rid . ".owl.gz";
if (!file_exists($reaction_file) || $this->GetParameterValue('download') == 'true') {
$url = $rest_uri . 'searchKineticLaws/biopax?q=SabioReactionID:' . $rid;
$data = file_get_contents($url);
if ($data === FALSE) {
continue;
}
$f = new FileFactory($reaction_file, true);
$f->Write($data);
$f->Close();
}
$buf = file_get_contents("compress.zlib://" . $reaction_file);
// send for parsing
$p = new BioPAX2Bio2RDF($this);
$p->SetBuffer($buf)->SetBioPAXVersion(3)->SetBaseNamespace("http://sabio.h-its.org/biopax#")->SetBio2RDFNamespace("http://bio2rdf.org/sabiork:")->SetDatasetURI($this->GetDatasetURI());
$rdf = $p->Parse();
parent::getWriteFile()->Write($rdf);
}
parent::getWriteFile()->Close();
//generate dataset description
echo "Generating dataset description... ";
$source_file = (new DataResource($this))->setURI("http://sabiork.h-its.org/sabioRestWebServices/searchKineticLaws/biopax")->setTitle("SABIO-RK Biochemical Reaction Kinetics Database")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($odir . $ofile)))->setFormat("text/xml")->setPublisher("http://sabio.villa-bosch.de/")->setHomepage("http://sabio.villa-bosch.de/")->setRights("use-share-modify")->setRights("no-commercial")->setLicense("http://sabio.villa-bosch.de/layouts/content/termscondition.gsp")->setDataset("http://identifiers.org/sabiork.reaction/");
$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}/")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/sabiork/sabiork.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() . $output_file->toRDF();
//write dataset description to file
parent::setGraphURI($graph_uri);
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
}
function Run()
{
$ldir = $this->GetParameterValue('indir');
$odir = $this->GetParameterValue('outdir');
$rfile = $this->GetParameterValue('download_url');
$lfile = substr($rfile, strrpos($rfile, "/") + 1);
// check if exists
if (!file_exists($ldir . $lfile) or parent::getParameterValue('download') == 'true') {
echo "dowloading {$rfile} ...";
trigger_error("Will attempt to download ", E_USER_NOTICE);
Utils::DownloadSingle($rfile, $ldir . $lfile);
echo "done" . PHP_EOL;
}
// make sure we have the zip archive
$zin = new ZipArchive();
if ($zin->open($ldir . $lfile) === FALSE) {
trigger_error("Unable to open {$ldir}{$lfile}");
exit;
}
// get the work
if ($this->GetParameterValue('files') == 'all') {
$files = explode("|", $this->GetParameterList('files'));
array_shift($files);
} else {
$files = explode("|", $this->GetParameterValue('files'));
}
$gz = false;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
$outfile = "ndc." . parent::getParameterValue('output_format');
parent::setWriteFile($odir . $outfile, $gz);
// now go through each item in the zip file and process
foreach ($files as $file) {
echo "Processing {$file}... ";
$fpin = $zin->getStream($file . ".txt");
if (!$fpin) {
trigger_error("Unable to get pointer to {$file} in {$ldir}{$lfile}", E_USER_ERROR);
return FALSE;
}
$this->{$file}($fpin);
parent::writeRDFBufferToWriteFile();
echo "done!" . PHP_EOL;
}
parent::getWriteFile()->close();
echo "Generating dataset description for {$outfile}... ";
//start generating dataset description file
$dataset_description = '';
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("FDA National Drug Code Directory")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($ldir . $lfile)))->setFormat("text/tab-separated-value")->setFormat("application/zip")->setPublisher("http://www.fda.gov")->setHomepage("http://www.fda.gov/Drugs/InformationOnDrugs/ucm142438.htm")->setRights("use-share")->setLicense(null)->setDataset("http://identifiers.org/ndc/");
$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}/{$outfile}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix}")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/ndc/ndc.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() . $output_file->toRDF();
//write dataset description to file
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
}
function run()
{
// get the file list
if ($this->GetParameterValue('files') == 'all') {
$files = explode("|", $this->GetParameterList('files'));
array_shift($files);
} else {
$files = explode(",", $this->GetParameterValue('files'));
}
if ($this->getParameterValue('additional') != 'none') {
$f = explode(",", $this->getParameterValue('additional'));
$files = array_merge($files, $f);
}
$ldir = $this->GetParameterValue('indir');
$odir = $this->GetParameterValue('outdir');
$rdir = $this->GetParameterValue('download_url');
$dataset_description = '';
foreach ($files as $file) {
$suffix = ".zip";
$lfile = $ldir . $file . $suffix;
$rfile = $rdir . $file . $suffix;
if ($file == "offsides" and !file_exists($lfile)) {
echo "downloading twosides...";
$rfile = "http://www.pharmgkb.org/redirect.jsp?p=ftp%3A%2F%2Fftpuserd%3AGKB4ftp%40ftp.pharmgkb.org%2Fdownload%2Ftatonetti%2F3003377s-offsides.zip";
utils::DownloadSingle($rfile, $lfile);
echo "done" . PHP_EOL;
} elseif ($file == "twosides" and !file_exists($lfile)) {
echo "downloading {$file} ...";
$rfile = "http://www.pharmgkb.org/redirect.jsp?p=ftp%3A%2F%2Fftpuserd%3AGKB4ftp%40ftp.pharmgkb.org%2Fdownload%2Ftatonetti%2F3003377s-twosides.zip";
utils::DownloadSingle($rfile, $lfile);
echo "done" . PHP_EOL;
} elseif ($file == 'annotations' or $file == 'relationships') {
if (!file_exists($lfile)) {
echo "Contact PharmGKB to get access to variants/clinical variants; save file as annotations.zip" . PHP_EOL;
continue;
}
} else {
if (!file_exists($lfile) or parent::getParameterValue('download') == true) {
echo "Downloading {$lfile} ... ";
Utils::DownloadSingle('https://www.pharmgkb.org/download.do?objId=' . $file . '.zip&dlCls=common', $lfile);
echo "done" . PHP_EOL;
}
}
// get a pointer to the file in the zip archive
if (!file_exists($lfile)) {
echo "no local copy of {$lfile} . skipping" . PHP_EOL;
continue;
}
$zin = new ZipArchive();
if ($zin->open($lfile) === FALSE) {
trigger_error("Unable to open {$lfile}");
exit;
}
$zipentries = array();
if ($file == "annotations") {
// exclude: 'clinical_ann.tsv','study_parameters.tsv'
$zipentries = array('clinical_ann_metadata.tsv', 'var_drug_ann.tsv', 'var_pheno_ann.tsv', 'var_fa_ann.tsv');
} else {
if ($file == "pathways") {
for ($i = 0; $i < $zin->numFiles; $i++) {
$stat = $zin->statIndex($i);
$entry = $stat['name'];
$ext = pathinfo($entry, PATHINFO_EXTENSION);
if ($ext != "txt") {
$zipentries[] = $entry;
}
}
} else {
if ($file == "relationships") {
$zipentries = array("relationships.tsv");
} else {
if ($file == 'offsides') {
$zipentries = array('3003377s-offsides.tsv');
} else {
if ($file == 'twosides') {
$zipentries = array('3003377s-twosides.tsv');
} else {
$zipentries = array($file . ".tsv");
}
}
}
}
}
// set the write file, parse, write and close
$suffix = parent::getParameterValue('output_format');
$outfile = $file . '.' . $suffix;
$gz = false;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
$this->SetWriteFile($odir . $outfile, $gz);
foreach ($zipentries as $zipentry) {
if (($fp = $zin->getStream($zipentry)) === FALSE) {
trigger_error("Unable to get {$file}.tsv in ziparchive {$lfile}");
return FALSE;
}
$this->SetReadFile($lfile);
$this->GetReadFile()->SetFilePointer($fp);
if ($file == "annotations") {
$fnx = substr($zipentry, 0, strpos($zipentry, ".tsv"));
echo "processing {$zipentry}..";
} else {
if ($file == 'pathways') {
$fnx = 'pathways';
echo "processing {$fnx} ({$zipentry})... ";
} else {
$fnx = $file;
echo "processing {$fnx} ... ";
}
}
$this->{$fnx}();
parent::writeRDFBufferToWriteFile();
parent::clear();
echo "done!" . PHP_EOL;
// generate the dataset release file
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("Pharmacogenomics Knowledge Base ({$zipentry})")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat("text/tab-separated-value")->setFormat("application/zip")->setPublisher("http://www.pharmgkb.org/")->setHomepage("http://www.pharmgkb.org/")->setRights("use")->setRights("no-commercial")->setLicense("http://www.pharmgkb.org/page/policies")->setDataset("http://identifiers.org/pharmgkb/");
$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}/{$outfile}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} {$file} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/pharmgkb/pharmgkb.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();
}
$this->GetWriteFile()->Close();
}
// foreach
echo "Generating dataset description... ";
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
}
function Run()
{
$file = "hgnc_complete_set.txt.gz";
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$rdir = parent::getParameterValue('download_url');
$lfile = $ldir . $file;
if (!file_exists($lfile) && parent::getParameterValue('download') == false) {
trigger_error($lfile . " not found. Will attempt to download.", E_USER_NOTICE);
parent::setParameterValue('download', true);
}
//download the hgnc file
$rfile = null;
if (parent::getParameterValue('download') == true) {
$rfile = $rdir;
echo "downloading {$file} ... ";
Utils::DownloadSingle($rfile, $lfile);
}
$ofile = $odir . "hgnc." . parent::getParameterValue('output_format');
$gz = false;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
parent::setWriteFile($ofile, $gz);
parent::setReadFile($lfile, true);
echo "processing {$file}... ";
$this->process();
echo "done!" . PHP_EOL;
//close write file
parent::getWriteFile()->close();
echo PHP_EOL;
// generate the dataset release file
echo "generating dataset release file... ";
$dataset_description = '';
$source_file = (new DataResource($this))->setURI($rdir)->setTitle('HUGO Gene Nomenclature Committee (HGNC)')->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat('text/tab-separated-value')->setFormat('application/zip')->setPublisher('http://www.genenames.org/')->setHomepage('http://www.genenames.org/data/gdlw_columndef.html')->setRights('use')->setRights('attribution')->setLicense('http://www.genenames.org/about/overview')->setDataset(parent::getDatasetURI());
$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}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/hgnc/hgnc.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());
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
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();
$this->setWriteFile($odir . $this->getBio2RDFReleaseFile());
$this->getWriteFile()->write($dataset_description);
$this->getWriteFile()->close();
}
public function run()
{
if (parent::getParameterValue('files') == 'all') {
$files = explode("|", parent::getParameterList('files'));
array_shift($files);
} else {
$files = explode(",", parent::getParameterValue('files'));
}
$release = parent::getParameterValue('release');
$releaseb = "WS249";
$remote_files = array("geneIDs" => "species/c_elegans/annotation/geneIDs/c_elegans.PRJNA13758." . $release . ".geneIDs.txt.gz", "functional_descriptions" => "species/c_elegans/annotation/functional_descriptions/c_elegans.PRJNA13758." . $release . ".functional_descriptions.txt.gz", "gene_interactions" => "species/c_elegans/annotation/gene_interactions/c_elegans.PRJNA13758." . $release . ".gene_interactions.txt.gz", "gene_associations" => "releases/current-production-release/ONTOLOGY/gene_association." . $releaseb . ".wb", "phenotype_associations" => "releases/current-production-release/ONTOLOGY/phenotype_association." . $releaseb . ".wb");
$local_files = array("geneIDs" => "wormbase." . parent::getParameterValue('release') . ".genes.txt.gz", "functional_descriptions" => "wormbase." . parent::getParameterValue('release') . ".functional_descriptions.txt.gz", "gene_interactions" => "wormbase." . parent::getParameterValue('release') . ".gene_interactions.txt.gz", "gene_associations" => "wormbase." . parent::getParameterValue('release') . ".gene_association.wb", "phenotype_associations" => "wormbase." . parent::getParameterValue('release') . ".phenotype_associations.wb");
$idir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$rdir = parent::getParameterValue('download_url');
$dataset_description = '';
$graph_uri = parent::getGraphURI();
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
foreach ($files as $file) {
$lfile = $idir . $local_files[$file];
$rfile = $rdir . $remote_files[$file];
if (!file_exists($lfile) or parent::getParameterValue('download') == true) {
trigger_error($lfile . " not found. Will attempt to download." . PHP_EOL, E_USER_WARNING);
echo "Downloading {$rfile}... ";
Utils::DownloadSingle($rfile, $lfile);
echo "done!" . PHP_EOL;
}
if (strstr($lfile, "gz")) {
parent::setReadFile($lfile, TRUE);
} else {
parent::setReadFile($lfile, FALSE);
}
$suffix = parent::getParameterValue('output_format');
$ofile = "wormbase." . $file . "." . $suffix;
$gz = strstr(parent::getParameterValue('output_format'), "gz") ? true : false;
parent::setWriteFile($odir . $ofile, $gz);
echo "Processing {$file}... ";
$fnx = $file;
$this->{$fnx}();
echo "done!" . PHP_EOL;
parent::getWriteFile()->close();
// generate the dataset release file
echo "Generating dataset description for {$ofile}... ";
// dataset description
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("WormBase Release " . parent::getParameterValue('release') . " subset ({$file})")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat("text/tab-separated-value")->setFormat("application/gzip")->setPublisher("http://wormbase.org/")->setHomepage("http://wormbase.org/")->setRights("use")->setRights("restricted-by-source-license")->setLicense("http://www.wormbase.org/about/policies")->setDataset("http://identifiers.org/wormbase/");
$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 {$prefix} - {$file}")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/wormbase/wormbase.php")->setCreateDate($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() . $output_file->toRDF();
echo "done!" . PHP_EOL;
}
parent::setGraphURI($graph_uri);
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
}
function Run()
{
// get the file list
if (parent::getParameterValue('files') == 'all') {
$files = array('all');
} else {
$files = explode(",", parent::getParameterValue('files'));
}
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$rdir = parent::getParameterValue('download_url');
$dataset_description = '';
foreach ($files as $file) {
$download = parent::getParameterValue('download');
$version = parent::getParameterValue("version");
$zip_file = ucfirst($file) . ".mitab." . $version . ".txt.zip";
$lfile = $ldir . $zip_file;
$gz = strstr(parent::getParameterValue('output_format'), ".gz") === FALSE ? false : true;
$ofile = "irefindex-" . $file . "." . parent::getParameterValue('output_format');
if (!file_exists($lfile)) {
trigger_error($lfile . " not found. Will attempt to download.", E_USER_NOTICE);
$download = true;
}
$rfile = $rdir . $zip_file;
if ($download == true) {
echo "downloading {$rfile}" . PHP_EOL;
if (FALSE === Utils::DownloadSingle($rfile, $lfile)) {
trigger_error("Error in Download");
return FALSE;
}
}
$zin = new ZipArchive();
if ($zin->open($lfile) === FALSE) {
trigger_error("Unable to open {$lfile}");
exit;
}
if ($zin->numFiles != 1) {
trigger_error("Found more than one file ... using first file");
}
$f = $zin->statIndex(0);
$base_file = $f['name'];
if (($fp = $zin->getStream($base_file)) === FALSE) {
trigger_error("Unable to get {$base_file} in ziparchive {$lfile}");
return FALSE;
}
parent::setReadFile($lfile);
parent::getReadFile()->setFilePointer($fp);
echo "Processing " . $file . " ...";
parent::setWriteFile($odir . $ofile, true);
if ($this->Parse() === FALSE) {
trigger_error("Parsing Error");
exit;
}
parent::writeRDFBufferToWriteFile();
parent::getWriteFile()->close();
$zin->close();
echo "Done!" . PHP_EOL;
$graph_uri = parent::getGraphURI();
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
// dataset description
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("iRefIndex ({$zip_file}")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat("text/tab-separated-value")->setFormat("application/zip")->setPublisher("http://irefindex.uio.no")->setHomepage("http://irefindex.uio.no")->setRights("use")->setRights("by-attribution")->setRights("no-commercial")->setLicense("http://irefindex.uio.no/wiki/README_MITAB2.6_for_iRefIndex#License")->setDataset("http://identifiers.org/irefindex/");
$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 {$prefix} - {$file}")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/irefindex/irefindex.php")->setCreateDate($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() . $output_file->toRDF();
parent::setGraphURI($graph_uri);
}
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
return TRUE;
}
/**
* process a results xml file from the download directory
**/
function process_file($infile)
{
$indir = parent::getParameterValue('indir');
$xml = new CXML($infile);
$this->setCheckPoint('file');
while ($xml->Parse("clinical_study") == TRUE) {
$this->setCheckPoint('record');
$this->root = $root = $xml->GetXMLRoot();
$this->nct_id = $nct_id = $this->getString("//id_info/nct_id");
$this->study_id = $study_id = parent::getNamespace() . "{$nct_id}";
### declare
$label = $this->getString("//brief_title");
if (!$label) {
$label = $this->getString("//official_title");
}
if (!$label) {
$label = "Clinical trial #" . $nct_id;
}
parent::addRDF(parent::describeIndividual($study_id, $label, parent::getVoc() . "Clinical-Study") . parent::describeClass(parent::getVoc() . "Clinical-Study", "Clinical Study"));
##########################################################################################
#required header
##########################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "download-date", $this->getString('//required_header/download_date')) . parent::triplify($study_id, parent::getVoc() . "url", $this->getString('//required_header/url')));
##########################################################################################
#identifiers
##########################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "nct-id", $this->getString('//id_info/nct_id'), "xsd:string") . parent::triplifyString($study_id, parent::getVoc() . "org-study-id", $this->getString('//id_info/org_study_id'), "xsd:string"));
$sids = $root->xpath('//id_info/secondary_id');
if (isset($sids)) {
foreach ($sids as $id) {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "secondary-id", (string) $id, "xsd:string"));
}
}
$nctaliases = $root->xpath('//id_info/nct-alias');
if (isset($nctaliases)) {
foreach ($nctaliases as $id) {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "nct-alias", (string) $id, "xsd:string"));
}
}
##########################################################################################
#titles
##########################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "brief-title", $this->getString("//brief_title")) . parent::triplifyString($study_id, parent::getVoc() . "official-title", $this->getString("//official_title")));
###################################################################################
#brief summary
###################################################################################
$brief_summary = str_replace(array("\r", "\n", "\t"), array("
", "
", "	"), $this->getString('//brief_summary/textblock'));
parent::addRDF(parent::triplifyString($study_id, $this->getVoc() . "brief-summary", $brief_summary));
####################################################################################
# detailed description
####################################################################################
$d = str_replace(array("\r", "\n", "\t"), array("
", "
", "	"), $this->getString('//detailed_description/textblock'));
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "detailed-description", $d));
#########################################################################################
#acronym
#########################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "acronym", $this->getString("//acronym")));
########################################################################################
#sponsors
########################################################################################
try {
$sponsors = array("lead_sponsor", "collaborator");
foreach ($sponsors as $sponsor) {
$a = @array_shift($root->xpath('//sponsors/' . $sponsor));
if ($a == null) {
break;
}
$agency = $this->getString("//agency", $a);
$agency_id = parent::getRes() . md5($agency);
$agency_class = $this->getString("//agency_class", $a);
$agency_class_id = parent::getRes() . md5($agency_class);
parent::addRDF(parent::describeIndividual($agency_id, $agency, parent::getVoc() . "Organization") . parent::describeClass(parent::getVoc() . "Organization", "Organization") . parent::triplify($study_id, parent::getVoc() . str_replace("_", "-", $sponsor), $agency_id) . parent::describeIndividual($agency_class_id, $agency_class, parent::getVoc() . "Organization") . parent::describeClass(parent::getVoc() . "Organization", "Organization") . parent::triplify($agency_id, parent::getVoc() . "organization", $agency_class_id));
}
} catch (Exception $e) {
echo "There was an error in the lead sponsor element: {$e}\n";
}
#################################################################################
# source
#################################################################################
$source = $this->getString('//source');
if ($source) {
$source_id = parent::getRes() . md5($source);
parent::addRDF(parent::describeIndividual($source_id, $source, parent::getVoc() . "Organization") . parent::triplify($study_id, parent::getVoc() . "source", $source_id));
}
######################################################################################
# oversight
######################################################################################
try {
$oversight = @array_shift($root->xpath('//oversight_info'));
$oversight_id = parent::getRes() . md5($oversight->asXML());
$authority = $this->getString('//authority', $oversight);
$authority_id = parent::getRes() . md5($authority);
parent::addRDF(parent::describeIndividual($oversight_id, $authority, parent::getVoc() . "Organization") . parent::triplify($study_id, $this->getVoc() . "oversight", $oversight_id) . parent::triplify($study_id, $this->getVoc() . "authority", $authority_id) . parent::triplifyString($oversight_id, parent::getVoc() . "has-dmc", $this->getString('//has_dmc', $oversight)));
} catch (Exception $e) {
echo "There was an error in the oversight info element: {$e}\n";
}
#################################################################################
# overall status
#################################################################################
$overall_status = $this->getString('//overall_status');
if ($overall_status) {
$status_id = parent::getRes() . md5($overall_status);
parent::addRDF(parent::describeIndividual($status_id, $overall_status, parent::getVoc() . "Status") . parent::describeClass(parent::getVoc() . "Status", "Status") . parent::triplify($study_id, parent::getVoc() . "overall-status", $status_id));
}
#########################################################################################
#why stopped
#########################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "why-stopped", $this->getString("//why_stopped")));
##################################################################################
# dates
##################################################################################
$dates = array("start_date", "end_date", "completion_date", "primary_completion_date", "verification_date", "lastchanged_date", "firstreceived_date", "firstreceived_results_date");
foreach ($dates as $date) {
$d = $this->getString('//' . $date);
if ($d) {
$datetime = $this->getDatetimeFromDate($d);
if (isset($datetime)) {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . str_replace("_", "-", $date), $datetime));
} else {
trigger_error("unable to parse date: {$d}", E_USER_ERROR);
}
}
}
####################################################################################
# phase
####################################################################################
$phase = $this->getString('//phase');
if ($phase && $phase != "N/A") {
$phase_id = $this->getRes() . md5($phase);
parent::addRDF(parent::describeIndividual($phase_id, $phase, parent::getVoc() . "Phase", $phase) . parent::describeClass(parent::getVoc() . "Phase", $phase) . parent::triplify($study_id, parent::getVoc() . "phase", $phase_id));
}
###################################################################################
# study type
####################################################################################
$study_type = $this->getString('//study_type');
if ($study_type) {
$study_type_id = $this->getRes() . md5($study_type);
parent::addRDF(parent::describeClass($study_type_id, $study_type, parent::getVoc() . "Study-Type") . parent::describeClass(parent::getVoc() . "Study-Type", "Study Type") . parent::triplify($study_id, parent::getVoc() . "study-type", $study_type_id));
}
###############################################################################
# study design
###############################################################################
$study_design = $this->getString('//study_design');
if ($study_design) {
$study_design_id = parent::getRes() . md5($study_id . $study_design);
parent::addRDF(parent::describeIndividual($study_design_id, "{$study_id} study design", parent::getVoc() . "Study-Design") . parent::describeClass(parent::getVoc() . "Study-Design", "Study Design") . parent::triplify($study_id, parent::getVoc() . "study-design", $study_design_id));
// Intervention Model: Parallel Assignment, Masking: Double-Blind, Primary Purpose: Treatment
foreach (explode(", ", $study_design) as $i => $b) {
$c = explode(": ", $b);
if (isset($c[1])) {
$sdp = $study_design_id . "-" . ($i + 1);
$key = parent::getRes() . md5($c[0]);
$value = parent::getRes() . md5($c[1]);
parent::addRDF(parent::describeIndividual($sdp, $b, parent::getVoc() . "Study-Design-Parameter") . parent::describeClass(parent::getVoc() . "Study-Design-Parameter", "Study Design Parameter") . parent::triplify($sdp, parent::getVoc() . "key", $key) . parent::describeClass($key, $c[0]) . parent::triplify($sdp, parent::getVoc() . "value", $value) . parent::describeClass($value, $c[1]) . parent::triplify($study_design_id, parent::getVoc() . "study-design-parameter", $sdp));
}
}
}
####################################################################################
# target duration
####################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "target-duration", $this->getString('//target_duration')));
################################################################################
# outcomes
###############################################################################
$outcomes = array("primary_outcome", "secondary_outcome", "other_outcome");
foreach ($outcomes as $outcome) {
$o = $root->xpath('//' . $outcome);
if ($o) {
$os = $o;
if (!is_array($o)) {
$os = array($o);
}
foreach ($os as $o) {
try {
$po_id = parent::getRes() . md5($nct_id . $o->asXML());
$po_type = parent::getVoc() . str_replace("_", "-", $outcome);
$measure = $this->getString('//measure', $o);
$time_frame = $this->getString('//time_frame', $o);
$safety_issue = $this->getString('//saftey_issue', $o);
$description = $this->getString('//description', $o);
parent::addRDF(parent::describeIndividual($po_id, $measure . " " . $time_frame, ucfirst($po_type)) . parent::describeClass(ucfirst($po_type), str_replace("_", " ", ucfirst($outcome))) . parent::triplifyString($po_id, "dc:description", $description) . parent::triplifyString($po_id, parent::getVoc() . "measure", $measure) . parent::triplifyString($po_id, parent::getVoc() . "time-frame", $time_frame) . parent::triplifyString($po_id, parent::getVoc() . "safety-issue", $safety_issue) . parent::triplify($study_id, parent::getVoc() . $po_type, $po_id));
} catch (Exception $e) {
echo "There was an error parsing the primary outcome element: {$e} \n";
}
}
}
}
##############################################################################
#number of arms
##############################################################################
try {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "number-of-arms", $this->getString('//number_of_arms')));
} catch (Exception $e) {
echo "There was an exception parsing the number of arms element: {$e}\n";
}
##############################################################################
#number of groups
##############################################################################
try {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "number-of-arms", $this->getString('//number_of_groups')));
} catch (Exception $e) {
echo "There was an exception parsing the number of groups: {$e}\n";
}
##############################################################################
#enrollment
##############################################################################
try {
$e = $root->xpath('//enrollment');
if ($e) {
$type = strtolower((string) $e[0]->attributes()->type);
$value = $this->getString('//enrollment');
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . ($type ? $type . "-" : "") . "enrollment", $value));
}
} catch (Exception $e) {
echo "There was an exception parsing the enrollment element: {$e}\n";
}
###############################################################################
#condition
###############################################################################
try {
$conditions = $root->xpath('//condition');
foreach ($conditions as $condition) {
$mesh_label_id = parent::getRes() . md5($condition);
parent::addRDF(parent::triplify($study_id, parent::getVoc() . "condition", $mesh_label_id) . parent::describeClass($mesh_label_id, $condition, parent::getVoc() . "Condition") . parent::describeClass(parent::getVoc() . "Condition", "Condition"));
}
} catch (Exception $e) {
echo "There was an exception parsing condition element: {$e}\n";
}
################################################################################
# arm_group
################################################################################
try {
$arm_groups = $root->xpath('//arm_group');
foreach ($arm_groups as $arm_group) {
$arm_group_id = $this->getString('./arm_group_label', $arm_group);
$arm_group_id = md5($arm_group_id);
$arm_group_uri = parent::getRes() . $this->nct_id . "/arm-group/" . $arm_group_id;
$arm_group_label = $this->nct_id . " arm group " . $arm_group_id;
$arm_group_type = ucfirst(str_replace(" ", "_", $this->getString('./arm_group_type', $arm_group)));
if (!$arm_group_type) {
$arm_group_type = "Clinical-Arm";
}
$description = $this->getString('./description', $arm_group);
parent::addRDF(parent::describeIndividual($arm_group_uri, $arm_group_label, parent::getVoc() . $arm_group_type) . parent::describeClass(parent::getVoc() . $arm_group_type, ucfirst(str_replace("_", " ", $arm_group_type))) . parent::triplifyString($arm_group_uri, parent::getVoc() . "description", $description) . parent::describeIndividual($arm_group_uri, $arm_group, parent::getVoc() . "Arm-Group") . parent::describeClass(parent::getVoc() . "Arm-Group", "Arm Group") . parent::triplify($study_id, parent::getVoc() . "arm-group", $arm_group_uri));
}
} catch (Exception $e) {
echo "There was an exception in arm groups: {$e}\n";
}
##############################################################################
#intervention
##############################################################################
try {
$interventions = $root->xpath('//intervention');
foreach ($interventions as $intervention) {
$intervention_id = parent::getRes() . md5($intervention->asXML());
$intervention_name = $this->getString('./intervention_name', $intervention);
$intervention_type = $this->getString('./intervention_type', $intervention);
$intervention_type_uri = parent::getVoc() . ucfirst(str_replace(" ", "_", $intervention_type));
$intervention_desc = $this->getString('./description', $intervention);
$intervention_on = $this->getString('./other_name', $intervention);
parent::addRDF(parent::describeIndividual($intervention_id, $intervention_name, $intervention_type_uri) . parent::describeClass($intervention_type_uri, $intervention_type) . parent::triplifyString($intervention_id, parent::getVoc() . "intervention-name", $intervention_name) . parent::triplifyString($intervention_id, parent::getVoc() . "intervention-desc", $intervention_desc) . parent::triplifyString($intervention_id, parent::getVoc() . "other-name", $intervention_on) . parent::triplify($study_id, parent::getvoc() . "intervention", $intervention_id));
$agl = $intervention->xpath("./arm_group_label");
foreach ($agl as $a) {
$arm_group_id = md5($a);
$ag = parent::getRes() . $this->nct_id . "/arm-group/" . $arm_group_id;
parent::addRDF(parent::describeIndividual($ag, $a, parent::getVoc() . "Arm-Group") . parent::describeClass(parent::getVoc() . "Arm-Group", "Arm Group") . parent::triplify($intervention_id, parent::getVoc() . "arm-group", $ag));
}
}
} catch (Exception $e) {
echo "There was an error in interventions {$e}\n";
}
###############################################################################
#eligibility
################################################################################
try {
$eligibility = @array_shift($root->xpath('//eligibility'));
if ($eligibility !== null) {
$eligibility_label = "eligibility for " . $study_id;
$eligibility_id = parent::getRes() . md5($eligibility->asXML());
parent::addRDF(parent::describeIndividual($eligibility_id, $eligibility_label, parent::getVoc() . "Eligibility") . parent::describeClass(parent::getVoc() . "Eligibility", "Eligibility") . parent::triplify($study_id, parent::getVoc() . "eligibility", $eligibility_id));
if ($criteria = @array_shift($eligibility->xpath('./criteria'))) {
$text = @array_shift($criteria->xpath('./textblock'));
parent::addRDF(parent::triplifyString($eligibility_id, parent::getVoc() . "text", $text));
$c = preg_split("/(Inclusion Criteria\\:|Exclusion Criteria\\:)/", $text);
//inclusion
if (isset($c[1])) {
$d = explode(" - ", $c[1]);
// the lists are separated by a hyphen
foreach ($d as $inclusion) {
$inc = trim($inclusion);
if ($inc != '') {
$inc_id = parent::getRes() . md5($inc);
parent::addRDF(parent::describeIndividual($inc_id, $inc, parent::getVoc() . "Inclusion-Criteria") . parent::describeClass(parent::getVoc() . "Inclusion-Criteria", "Inclusion Criteria") . parent::triplify($eligibility_id, parent::getVoc() . "inclusion-criteria", $inc_id));
}
}
}
//exclusion
if (isset($c[2])) {
$d = explode(" - ", $c[1]);
foreach ($d as $exclusion) {
$exc = trim($exclusion);
if ($exc != '') {
$exc_id = parent::getRes() . md5($exc);
parent::addRDF(parent::describeIndividual($exc_id, $exc, parent::getVoc() . "Exclusion-Criteria") . parent::describeClass(parent::getVoc() . "Exclusion-Criteria", "Exclusion Criteria") . parent::triplify($eligibility_id, parent::getVoc() . "exclusion-criteria", $exc_id));
}
}
}
}
parent::addRDF(parent::triplifyString($eligibility_id, parent::getVoc() . "gender", $this->getString('./gender', $eligibility)));
parent::addRDF(parent::triplifyString($eligibility_id, parent::getVoc() . "healthy-volunteers", $this->getString('./healthy_volunteers', $eligibility)));
$attributes = array('minimum_age', 'maximum_age');
foreach ($attributes as $a) {
$s = $this->getString('./' . $a, $eligibility);
if ($s != 'N/A') {
$age = trim(str_replace("Years", "", $s));
parent::addRDF(parent::triplifyString($eligibility_id, parent::getVoc() . str_replace("_", "-", $a), $age));
}
}
$attributes = array("study_pop" => "study-population", "sampling_method" => "sampling-method");
foreach ($attributes as $a => $r) {
$e = @array_shift($eligibility->xpath('./' . $a));
if ($s = $this->getString('./' . $a, $eligibility)) {
parent::addRDF(parent::triplifyString($eligibility_id, parent::getVoc() . $r, $this->getString('./textblock', $e)));
}
}
}
} catch (Exception $e) {
echo "There was an error in eligibility: {$e}\n";
}
######################################################################################
#biospec
#####################################################################################
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "biospec-retention", $this->getString('//biospec_retention')));
try {
$b = @array_shift($root->xpath('//biospec_descr'));
if ($b) {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "biospec_descr", $this->getString('./textblock', $b)));
}
} catch (Exception $e) {
echo "There was an error in biospec_descr: {$e}\n";
}
###################################################################
# contacts
###################################################################
$contacts = array("overall_official", "overall_contact", "overall_contact_backup");
try {
foreach ($contacts as $c) {
$d = @array_shift($root->xpath('//' . $c));
if ($d) {
parent::addRDF(parent::triplify($study_id, parent::getVoc() . str_replace("_", "-", $c), $this->makeContact($d)));
}
}
} catch (Exception $e) {
echo "There was an error parsing overall contact: {$e}" . "\n";
}
##############################################################
# location of facility doing the testing
##############################################################
try {
$location = @array_shift($root->xpath('//location'));
if ($location) {
$location_uri = parent::getRes() . md5($location->asXML());
$name = $this->getString('//facility/name', $location);
$address = @array_shift($location->xpath('//facility/address'));
$contact = @array_shift($location->xpath('//contact'));
$backups = @array_shift($location->xpath('//contact_backup'));
$investigators = @array_shift($location->xpath('//investigator'));
parent::addRDF(parent::describeIndividual($location_uri, $name, parent::getVoc() . "Location") . parent::describeClass(parent::getVoc() . "Location", "Location") . parent::triplifyString($location_uri, parent::getVoc() . "status", $this->getString('//status', $location)) . parent::triplify($study_id, parent::getVoc() . "location", $location_uri) . parent::triplify($location_uri, parent::getVoc() . "address", $this->makeAddress($address)) . ($contact != null ? parent::triplify($location_uri, parent::getVoc() . "contact", $this->makeContact($contact)) : ""));
if ($backups) {
foreach ($backups as $backup) {
parent::addRDF(parent::triplify($location_uri, parent::getVoc() . "contact-backup", $this->makeContact($backup)));
}
}
if ($investigators) {
foreach ($investigators as $investigator) {
parent::addRDF(parent::triplify($location_uri, parent::getVoc() . "investigator", $this->makeContact($investigator)));
}
}
}
} catch (Exception $e) {
echo "There was an error parsing location: {$e}" . "\n";
}
######################################################################
#countries
######################################################################
try {
$a = array("location_countries", "removed_countries");
foreach ($a as $country) {
$lc = @array_shift($root->xpath('//' . $country));
if ($lc) {
$label = $this->getString('//country', $lc);
$cid = parent::getRes() . md5($label);
parent::addRDF(parent::describeIndividual($cid, $label, parent::getVoc() . "Country") . parent::describeClass(parent::getVoc() . "Country", "Country") . parent::triplify($study_id, parent::getVoc() . "country", $cid));
}
}
} catch (Exception $e) {
echo "There was an error parsing country: {$e}" . "\n";
}
######################################################################
#reference
######################################################################
try {
$a = array("reference", "result_reference");
foreach ($a as $ref_type) {
$references = $root->xpath('//' . $ref_type);
foreach ($references as $reference) {
$p = $this->getString('./PMID', $reference);
if ($p) {
$pmid = "pubmed:{$p}";
parent::addRDF(parent::describeIndividual($pmid, $p, parent::getVoc() . "Reference") . parent::describeClass(parent::getVoc() . "Reference", "Reference") . parent::triplifyString($pmid, parent::getVoc() . "citation", $this->getString('./citation', $reference)) . parent::triplify($study_id, parent::getVoc() . str_replace("_", "-", $ref_type), $pmid));
}
}
}
} catch (Exception $e) {
echo "There was an error parsing references element: {$e}\n";
}
#######################################################################
#link
#######################################################################
try {
$links = $root->xpath('//link');
foreach ($links as $i => $link) {
$url = $this->getString('./url', $link);
$url = preg_replace("/>.*\$/", "", $url);
$lid = parent::getRes() . md5($url);
parent::addRDF(parent::describeIndividual($lid, $this->getString('./description', $link), parent::getVoc() . "Link") . parent::describeClass(parent::getVoc() . "Link", "Link") . parent::triplify($lid, parent::getVoc() . "url", $url) . parent::triplify($study_id, parent::getVoc() . "link", $lid));
}
} catch (Exception $e) {
echo "There was an error parsing link element: {$e}\n";
}
############################################################################
#responsible party
############################################################################
try {
$rp = @array_shift($root->xpath('//responsible_party'));
if ($rp) {
$rp_id = parent::getRes() . md5($rp->asXML());
$label = $this->getString('./name_title', $rp);
if (!$label) {
$label = $this->getString('./organization', $rp);
} else {
$label .= ", " . $this->getString('./organization', $rp);
}
if (!$label) {
$label = $this->getString('./party_type', $rp);
}
$org_id = parent::getRes() . md5($this->getString('./organization', $rp));
parent::addRDF(parent::describeIndividual($rp_id, $label, parent::getVoc() . "Responsible-Party") . parent::describeClass(parent::getVoc() . "Responsible-Party", "Responsible Party") . parent::triplify($study_id, parent::getVoc() . "responsible-party", $rp_id) . parent::triplify($rp_id, parent::getVoc() . "organization", $org_id) . parent::describeIndividual($org_id, $this->getString('./organization', $rp), parent::getVoc() . "Organization") . parent::describeClass(parent::getVoc() . "Organization", "Organization") . parent::triplifyString($rp_id, parent::getVoc() . "name-title", $this->getString('./name_title', $rp)) . parent::triplifyString($rp_id, parent::getVoc() . "party-type", $this->getString('./party_type', $rp)) . parent::triplifyString($rp_id, parent::getVoc() . "investigator-affiliation", $this->getString('./investigator_affiliation', $rp)) . parent::triplifyString($rp_id, parent::getVoc() . "investigator-full-name", $this->getString('./investigator_full_name', $rp)) . parent::triplifyString($rp_id, parent::getVoc() . "investigator-title", $this->getString('./investigator_title', $rp)));
}
} catch (Exception $e) {
echo "There was an error parsing the responsible_party element: {$e}\n";
}
##############################################################################
# keywords
##############################################################################
try {
$keywords = $root->xpath('//keyword');
foreach ($keywords as $keyword) {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "keyword", (string) $keyword));
}
} catch (Exception $e) {
echo "There was an error parsing the keywords element: {$e}";
}
# mesh terms
# note: mesh terms are assigned using an imperfect algorithm
try {
$mesh_terms = $root->xpath('//condition_browse/mesh_term');
foreach ($mesh_terms as $mesh_term) {
$term = (string) $mesh_term;
$mesh_id = parent::getRes() . md5($term);
parent::addRDF(parent::triplify($study_id, parent::getVoc() . "condition-mesh", $mesh_id));
parent::addRDF(parent::triplifyString($mesh_id, "rdfs:label", $term));
}
} catch (Exception $e) {
echo "There was an error in mesh_terms: {$e}\n";
}
################################################################################
# regulated by fda? is section 801? has expanded access?
################################################################################
try {
parent::addRDF(parent::triplifyString($study_id, parent::getVoc() . "is-fda-regulated", $this->getString('is_fda_regulated')) . parent::triplifyString($study_id, parent::getVoc() . "is-section-801", $this->getString('is_section_801')) . parent::triplifyString($study_id, parent::getVoc() . "has-expanded-access", $this->getString('has_expanded_access')));
} catch (Exception $e) {
echo "There was an error parsing the is_fda_regulated element: {$e}\n";
}
###############################################################################
# mesh terms for the intervention browse
###############################################################################
try {
$a = array("condition_browse", "intervention_browse");
foreach ($a as $browse_type) {
$terms = $root->xpath("//{$browse_type}/mesh_term");
foreach ($terms as $term) {
$term_label = (string) $term;
$term_id = parent::getRes() . md5($term);
parent::addRDF(parent::describeIndividual($term_id, $term_label, parent::getVoc() . "Term") . parent::describeClass(parent::getVoc() . "Term", "Term") . parent::triplify($study_id, parent::getVoc() . str_replace("_", "-", $browse_type), $term_id));
}
}
} catch (Exception $e) {
echo "There was an error parsing {$browse_type}/mesh_term element: {$e}\n";
}
################################################################################
# clinical results
################################################################################
try {
$cr = @array_shift($root->xpath('//clinical_results'));
if ($cr) {
$cr_id = parent::getRes() . md5($study_id . $cr->asXML());
parent::addRDF(parent::describeIndividual($cr_id, "clinical results for {$study_id}", parent::getVoc() . "Clinical-Result") . parent::describeClass(parent::getVoc() . "Clinical-Result", "Clinical Result") . parent::triplifyString($cr_id, parent::getVoc() . "description", $this->getString('./desc', $cr)) . parent::triplifyString($cr_id, parent::getVoc() . "restrictive-agreement", $this->getString('./restrictive_agreement', $cr)) . parent::triplifyString($cr_id, parent::getVoc() . "limitations-and-caveats", $this->getString('./limitations_and_caveats', $cr)) . parent::triplify($study_id, parent::getVoc() . "clinical-result", $cr_id));
}
} catch (Exception $e) {
echo "There was an error parsing clinical results: {$e}\n";
}
################################################################################
# Participant Flow
################################################################################
try {
$pc = 1;
$mc = 1;
$wc = 1;
$pf = @array_shift($root->xpath('//clinical_results/participant_flow'));
if ($pf) {
$pf_id = parent::getRes() . md5($pf->asXML());
parent::addRDF(parent::describeIndividual($pf_id, "participant flow for {$study_id}", parent::getVoc() . "Participant-Flow") . parent::describeClass(parent::getVoc() . "Participant-Flow", "Participant-Flow") . parent::triplify($study_id, parent::getVoc() . "participant-flow", $pf_id) . parent::triplifyString($pf_id, parent::getVoc() . "recruitment-details", $this->getString('./recruitment_details', $pf)) . parent::triplifyString($pf_id, parent::getVoc() . "pre-assignment-details", $this->getString('./pre_assignment_details', $pf)));
$groups = @array_shift($pf->xpath('./group_list'));
foreach ($groups as $group) {
parent::addRDF(parent::triplify($pf_id, parent::getVoc() . "group", $this->makeGroup($group)));
}
//period_list
$periods = @array_shift($pf->xpath('./period_list'));
foreach ($periods as $period) {
$period_id = parent::getRes() . $nct_id . "/period/" . $pc++;
$period_title = $this->getString('./title', $period);
parent::addRDF(parent::describeIndividual($period_id, $period_title . " for {$nct_id}", parent::getVoc() . "Period") . parent::describeClass(parent::getVoc() . "Period", "Period") . parent::triplify($pf_id, parent::getVoc() . "period", $period_id));
// milestones
$milestones = @array_shift($period->xpath('./milestone_list'));
if ($milestones) {
foreach ($milestones as $milestone) {
$milestone_id = parent::getRes() . $nct_id . "/milestone/" . $mc++;
$label = $this->getString('./title', $milestone);
parent::addRDF(parent::describeIndividual($milestone_id, $label, parent::getVoc() . "Milestone") . parent::describeClass(parent::getVoc() . "Milestone", "Milestone") . parent::triplify($period_id, parent::getVoc() . "milestone", $milestone_id));
// participants
$p = 1;
$ps_list = @array_shift($milestone->xpath('./participants_list'));
foreach ($ps_list as $ps) {
$ps_id = $milestone_id . "/p/" . $p++;
$group_id = parent::getRes() . $this->nct_id . "/group/" . $ps->attributes()->group_id;
$count = (string) $ps->attributes()->count;
parent::addRDF(parent::describeIndividual($ps_id, "participant counts in " . $ps->attributes()->group_id . " for milestone {$mc} of {$nct_id}", parent::getVoc() . "Participant-Count") . parent::describeClass(parent::getVoc() . "Participant-Count", "Participant Count") . parent::triplify($ps_id, parent::getVoc() . "group", $group_id) . parent::triplifyString($ps_id, parent::getVoc() . "count", $count) . parent::triplify($milestone_id, parent::getVoc() . "participant-counts", $ps_id));
}
}
}
// milestones
$withdraws = @array_shift($period->xpath('./drop_withdraw_reason_list'));
if ($withdraws) {
foreach ($withdraws as $withdraw) {
$wid = parent::getRes() . $this->nct_id . "/withdraw/" . $wc++;
$label = $this->getString('./title', $withdraw);
parent::addRDF(parent::describeIndividual($wid, $label, parent::getVoc() . "Withdraw-Reason") . parent::describeClass(parent::getVoc() . "Withdraw-Reason", "Withdraw Reason"));
// participants
$ps_list = @array_shift($withdraw->xpath('./participants_list'));
foreach ($ps_list as $ps) {
$group_id = parent::getRes() . $nct_id . "/group/" . $ps->attributes()->group_id;
$count = (string) $ps->attributes()->count;
parent::addRDF(parent::triplify($wid, parent::getVoc() . "group", $group_id) . parent::triplifyString($wid, parent::getVoc() . "count", $count));
}
}
}
}
}
} catch (Exception $e) {
echo "There was an error parsing participant flow element: {$e}\n";
}
################################################################################
# baseline
################################################################################
try {
$baseline = @array_shift($root->xpath('//baseline'));
if ($baseline) {
$b_id = $this->nct_id . "/baseline";
$b_uri = parent::getRes() . $b_id;
// group list
$groups = @array_shift($baseline->xpath('./group_list'));
foreach ($groups as $group) {
parent::addRDF(parent::describeIndividual($b_uri, "baseline for {$nct_id}", parent::getVoc() . "Baseline") . parent::describeClass(parent::getVoc() . "Baseline", "Baseline") . parent::triplify($b_uri, parent::getVoc() . "group", $this->makeGroup($group)) . parent::triplify($study_id, parent::getVoc() . "baseline", $b_uri));
}
// measure list
$measures = @array_shift($baseline->xpath('./measure_list'));
foreach ($measures as $measure) {
parent::addRDF(parent::triplify($b_uri, parent::getVoc() . "measure", $this->makeMeasure($measure)));
}
}
} catch (Exception $e) {
echo "Error in parsing baseline" . PHP_EOL;
}
################################################################################
# outcomes
################################################################################
try {
$outcomes = @array_shift($root->xpath('//outcome_list'));
if ($outcomes) {
foreach ($outcomes as $i => $outcome) {
$outcome_id = $this->nct_id . "/outcome/" . ($i + 1);
$outcome_uri = parent::getRes() . $outcome_id;
$outcome_label = $this->getString("./title", $outcome);
if (!$outcome_label) {
$outcome_label = "outcome for " . $this->nct_id;
}
parent::addRDF(parent::describeIndividual($outcome_uri, $outcome_label, parent::getVoc() . "Outcome", $this->getString("./description", $outcome)) . parent::describeClass(parent::getVoc() . "Outcome", "Outcome") . parent::triplify($study_id, parent::getVoc() . "outcome", $outcome_uri) . parent::triplifyString($outcome_uri, parent::getVoc() . "type", $this->getString("./type", $outcome)) . parent::triplifyString($outcome_uri, parent::getVoc() . "time-frame", $this->getString("./time_frame", $outcome)) . parent::triplifyString($outcome_uri, parent::getVoc() . "safety-issue", $this->getString("./safety_issue", $outcome)) . parent::triplifyString($outcome_uri, parent::getVoc() . "posting-date", $this->getString("./posting-date", $outcome)) . parent::triplifyString($outcome_uri, parent::getVoc() . "population", $this->getString("./population", $outcome)));
$groups = @array_shift($outcome->xpath('./group_list'));
if ($groups) {
foreach ($groups as $group) {
parent::addRDF(parent::triplify($outcome_uri, parent::getVoc() . "group", $this->makeGroup($group)));
}
}
// measure list
$measures = @array_shift($outcome->xpath('./measure_list'));
if ($measures) {
foreach ($measures as $measure) {
parent::addRDF(parent::triplify($outcome_uri, parent::getVoc() . "measure", $this->makeMeasure($measure)));
}
}
// analysis list
$analyses = @array_shift($outcome->xpath('./analysis_list'));
if ($analyses) {
foreach ($analyses as $analysis) {
parent::addRDF(parent::triplify($outcome_uri, parent::getVoc() . "analysis", $this->makeAnalysis($analysis)));
}
}
}
}
} catch (Exception $e) {
echo "Error in parsing outcomes" . PHP_EOL;
}
################################################################################
# events
################################################################################
try {
$c_ev = $c_c = 1;
$reported_events = @array_shift($root->xpath('//reported_events'));
if ($reported_events) {
$rp_id = parent::getRes() . md5($reported_events->asXML());
$groups = @array_shift($reported_events->xpath('./group_list'));
parent::addRDF(parent::describeIndividual($rp_id, "Reported events for {$nct_id}", parent::getVoc() . "Reported-Events") . parent::describeClass(parent::getVoc() . "Reported-Events", "Reported Events") . parent::triplify($study_id, parent::getVoc() . "reported-events", $rp_id));
foreach ($groups as $group) {
parent::addRDF(parent::triplify($rp_id, parent::getVoc() . "group", $this->makeGroup($group)));
}
// events
$event_list = array("serious_events" => "Serious Event", "other_events" => "Other Event");
foreach ($event_list as $ev => $ev_label) {
$et = @array_shift($reported_events->xpath('./' . $ev));
if (!$et) {
continue;
}
$ev_uri = parent::getVoc() . str_replace(" ", "-", $ev_label);
$categories = @array_shift($et->xpath('./category_list'));
foreach ($categories as $category) {
$major_title = $this->getString('./title', $category);
$major_title_uri = parent::getRes() . md5($major_title);
$events = @array_shift($category->xpath('./event_list'));
foreach ($events as $event) {
$e_uri = parent::getRes() . $this->nct_id . "/{$ev}/" . $c_ev++;
$subtitle = (string) $this->getString('./sub_title', $event) . " for " . $this->nct_id;
$subtitle_uri = parent::getRes() . md5($subtitle);
parent::addRDF(parent::describeIndividual($e_uri, $subtitle, $ev_uri) . parent::describeClass($ev_uri, $ev_label) . parent::triplify($e_uri, parent::getVoc() . "sub-title", $subtitle_uri) . parent::describeIndividual($subtitle_uri, $subtitle, parent::getVoc() . "Event") . parent::describeClass(parent::getVoc() . "Event", "Event") . parent::triplify($e_uri, parent::getVoc() . "major-title", $major_title_uri) . parent::describeClass($major_title_uri, $major_title) . parent::triplify($rp_id, parent::getVoc() . str_replace("_", "-", $ev), $e_uri));
$counts = $event->xpath('./counts');
foreach ($counts as $c) {
$group_id = $c->attributes()->group_id;
$group_uri = parent::getRes() . $nct_id . "/group/" . $group_id;
$c_uri = $e_uri . "/count/" . $c_c++;
parent::addRDF(parent::describeIndividual($c_uri, $subtitle . " for " . $group_id . " in " . $this->nct_id, parent::getVoc() . "Event-Count") . parent::describeClass(parent::getVoc() . "Event-Count", "Event Count") . parent::triplify($c_uri, parent::getVoc() . "group", $group_uri) . parent::triplify($e_uri, parent::getVoc() . "count", $c_uri) . parent::triplifyString($c_uri, parent::getVoc() . "default-vocabulary", $this->getString('./default_vocab', $et)) . parent::triplifyString($c_uri, parent::getVoc() . "frequency-threshold", $this->getString('./frequency_threshold', $et)) . parent::triplifyString($c_uri, parent::getVoc() . "default-assessment", $this->getString('./default_assessment', $et)) . parent::triplifyString($c_uri, parent::getVoc() . "number-events", $c->attributes()->events) . parent::triplifyString($c_uri, parent::getVoc() . "subjects-affected", $c->attributes()->subjects_affected) . parent::triplifyString($c_uri, parent::getVoc() . "subjects-at-risk", $c->attributes()->subjects_at_risk));
}
}
}
}
}
} catch (Exception $e) {
echo "Error in parsing reported events" . PHP_EOL;
}
parent::writeRDFBufferToWriteFile();
}
$this->setCheckPoint('record');
$this->setCheckPoint('dataset');
}
private function process()
{
$z = 0;
$y = 1;
while ($l = $this->getReadFile()->Read(200000)) {
if ($z++ % 1000000 == 0) {
echo $z . PHP_EOL;
$odir = parent::getParameterValue('outdir');
$ofile = 'iproclass.' . $y++ . "." . parent::getParameterValue('output_format');
$gz = strstr(parent::getParameterValue('output_format'), "gz") ? true : false;
if (parent::getWriteFile() != null) {
parent::getWriteFile()->close();
parent::clear();
}
// generate a new file
parent::setWriteFile($odir . $ofile, $gz);
}
$fields = explode("\t", $l);
@($uniprot_acc = $fields[0]);
@($uniprot = $fields[1]);
@($gene = $fields[2]);
@($refseq = $fields[3]);
@($gi = $fields[4]);
@($pdb = $fields[5]);
@($pfam = $fields[6]);
@($go = $fields[7]);
@($pirsf = $fields[8]);
@($ipi = $fields[9]);
@($uniref_100 = $fields[10]);
@($uniref_90 = $fields[11]);
@($uniref_50 = $fields[12]);
@($uniparc = $fields[13]);
//skipping pir-psd because db no longer maintained
@($ncbi_taxonomy = $fields[15]);
@($mim = $fields[16]);
@($unigene = $fields[17]);
@($ensembl = $fields[18]);
@($pubmed = $fields[19]);
@($embl_genbank_ddbj = $fields[20]);
@($embl_protein = trim($fields[21]));
$id = $uniprot_acc;
$id_res = $this->getNamespace() . $id;
$id_label = "iproclass entry for uniprot:{$uniprot_acc}";
parent::addRDF(parent::triplify($id_res, $this->getVoc() . "x-uniprot", "uniprot:" . $uniprot_acc));
if (!empty($uniprot)) {
$uniprot_ids = explode("; ", $uniprot);
foreach ($uniprot_ids as $uniprot_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-uniprot", "uniprot:" . $uniprot_id));
}
}
if (!empty($gene)) {
$gene_ids = explode("; ", $gene);
foreach ($gene_ids as $gene_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-ncbigene", "geneid:" . $gene_id));
}
}
if (!empty($refseq)) {
$refseq_ids = explode("; ", $refseq);
foreach ($refseq_ids as $refseq_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-refseq", "refseq:" . $refseq_id));
}
}
if (!empty($gi)) {
$gi_ids = explode("; ", $gi);
foreach ($gi_ids as $gi_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-gi", "gi:" . $gi_id));
}
}
if (!empty($pdb)) {
$pdb_ids = explode("; ", $pdb);
foreach ($pdb_ids as $pdb_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-pdb", "pdb:" . $pdb_id));
}
}
if (!empty($pfam)) {
$pfam_ids = explode("; ", $pfam);
foreach ($pfam_ids as $pfam_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-pfam", "pfam:" . $pfam_id));
}
}
if (!empty($go)) {
$go_ids = explode("; ", $go);
foreach ($go_ids as $go_id) {
$go_id = substr($go_id, 3);
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-go", "go:" . $go_id));
}
}
if (!empty($pirsf)) {
$pirsf_ids = explode("; ", $pirsf);
foreach ($pirsf_ids as $pirsf_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-pirsf", "pirsf:" . $pirsf_id));
}
}
if (!empty($ipi)) {
$ipi_ids = explode("; ", $ipi);
foreach ($ipi_ids as $ipi_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-ipi", "ipi:" . $ipi_id));
}
}
if (!empty($uniref_100)) {
$uniref_100_ids = explode("; ", $uniref_100);
foreach ($uniref_100_ids as $uniref_100_id) {
parent::AddRDF(parent::QQuadO_URL($id_res, "rdfs:seeAlso", "http://uniprot.org/uniref/" . $uniref_100_id));
}
}
if (!empty($uniref_90)) {
$uniref_90_ids = explode("; ", $uniref_90);
foreach ($uniref_90_ids as $uniref_90_id) {
parent::AddRDF(parent::QQuadO_URL($id_res, "rdfs:seeAlso", "http://uniprot.org/uniref/" . $uniref_90_id));
}
}
if (!empty($uniref_50)) {
$uniref_50_ids = explode("; ", $uniref_50);
foreach ($uniref_50_ids as $uniref_50_id) {
parent::AddRDF(parent::QQuadO_URL($id_res, "rdfs:seeAlso", "http://uniprot.org/uniref/" . $uniref_50_id));
}
}
if (!empty($uniparc)) {
$uniparc_ids = explode("; ", $uniparc);
foreach ($uniparc_ids as $uniparc_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-uniparc", "uniparc:" . $uniparc_id) . parent::QQuadO_URL($id_res, "rdfs:seeAlso", "http://uniprot.org/uniparc/" . $uniparc_id));
}
}
if (!empty($ncbi_taxonomy)) {
$taxonomy_ids = explode("; ", $ncbi_taxonomy);
foreach ($taxonomy_ids as $taxonomy_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-taxon", "taxon:" . $taxonomy_id));
}
}
if (!empty($mim)) {
$mim_ids = explode("; ", $mim);
foreach ($mim_ids as $mim_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-omim", "omim:" . $mim_id));
}
}
if (!empty($unigene)) {
$unigene_ids = explode("; ", $unigene);
foreach ($unigene_ids as $unigene_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-unigene", "unigene:" . $unigene_id));
}
}
if (!empty($ensembl)) {
$ensembl_ids = explode("; ", $ensembl);
foreach ($ensembl_ids as $ensembl_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-ensembl", "ensembl:" . $ensembl_id));
}
}
if (!empty($pubmed)) {
$pubmed_ids = explode("; ", $pubmed);
foreach ($pubmed_ids as $pubmed_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-pubmed", "pubmed:" . $pubmed_id));
}
}
if (!empty($embl_genbank_ddbj)) {
$genbank_ids = explode("; ", $embl_genbank_ddbj);
foreach ($genbank_ids as $genbank_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-genbank", "genbank:" . $genbank_id));
}
}
if (!empty($embl_protein)) {
$embl_protein_ids = explode(";", $embl_protein);
foreach ($embl_protein_ids as $embl_protein_id) {
parent::AddRDF(parent::triplify($id_res, $this->getVoc() . "x-genbank", "genbank:" . $embl_protein_id));
}
}
//write rdf to file
$this->WriteRDFBufferToWriteFile();
}
//while
}
function Run()
{
// directory shortcuts
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
// get the work specified
$list = trim(parent::getParameterValue('files'));
if ($list == 'all') {
// call the getAllModelsId webservice
$file = $ldir . "all_models.json";
if (!file_exists($file)) {
try {
$x = @new SoapClient("http://www.ebi.ac.uk/biomodels-main/services/BioModelsWebServices?wsdl");
} catch (Exception $e) {
echo $e->getMessage();
}
$entries = $x->getAllModelsId();
file_put_contents($file, json_encode($entries));
} else {
$entries = json_decode(file_get_contents($file));
}
} elseif ($list == 'curated') {
// call the getAllCuratedModelsId webservice
$file = $ldir . "curated_models.json";
if (!file_exists($file)) {
try {
$x = @new SoapClient("http://www.ebi.ac.uk/biomodels-main/services/BioModelsWebServices?wsdl");
} catch (Exception $e) {
echo $e->getMessage();
}
$entries = $x->getAllCuratedModelsId();
file_put_contents($file, json_encode($entries));
} else {
$entries = json_decode(file_get_contents($file));
}
} else {
// check if a hyphenated list was provided
if (($pos = strpos($list, "-")) !== FALSE) {
$start_range = substr($list, 0, $pos);
$end_range = substr($list, $pos + 1);
for ($i = $start_range; $i <= $end_range; $i++) {
$entries[] = "BIOMD" . str_pad($i, 10, "0", STR_PAD_LEFT);
}
} else {
// for comma separated list
$b = explode(",", $this->GetParameterValue('files'));
foreach ($b as $e) {
$entries[] = "BIOMD" . str_pad($e, 10, "0", STR_PAD_LEFT);
}
}
}
$graph_uri = parent::getGraphURI();
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
// set the write file
$suffix = parent::getParameterValue('output_format');
$outfile = 'biomodels' . '.' . $suffix;
$gz = false;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
$dataset_description = '';
parent::setWriteFile($odir . $outfile, $gz);
// iterate over the entries
$i = 0;
$total = count($entries);
foreach ($entries as $id) {
echo "processing " . ++$i . " of {$total} - biomodel# " . $id;
$download_file = $ldir . $id . ".owl.gz";
$url = parent::getParameterValue('download_url') . "publ/{$id}/{$id}-biopax3.owl";
// download if the file doesn't exist or we are told to
if (!file_exists($download_file) || $this->GetParameterValue('download') == 'true') {
// download
echo " - downloading";
$ret = utils::downloadsingle($url, 'compress.zlib://' . $download_file, true);
if ($ret === false) {
echo "\nTrying non-curated model";
$url = parent::getParametervalue('download_url') . "uncura_publ/{$id}/{$id}-biopax3.owl";
$ret = utils::downloadsingle($url, 'compress.zlib://' . $download_file, true);
if ($ret === false) {
continue;
}
}
echo " - downloaded";
}
// load entry, parse and write to file
echo " - parsing... ";
// $this->SetReadFile($download_file,true);
$buf = file_get_contents("compress.zlib://" . $download_file);
$converter = new BioPAX2Bio2RDF($this);
$converter->SetBuffer($buf)->SetBioPAXVersion(3)->SetBaseNamespace("http://identifiers.org/biomodels.db/{$id}/")->SetBio2RDFNamespace("http://bio2rdf.org/biomodels:" . $id . "_")->SetDatasetURI($this->GetDatasetURI());
$rdf = $converter->Parse();
parent::addRDF($rdf);
parent::writeRDFBufferToWriteFile();
//generate dataset description
$source_file = (new DataResource($this))->setURI($url)->setTitle("EBI BioModels Database - BioModel # {$id}")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($download_file)))->setFormat("rdf/xml")->setPublisher("http://www.ebi.ac.uk/")->setHomepage("http://www.ebi.ac.uk/biomodels-main/")->setRights("use-share-modify")->setLicense("http://www.ebi.ac.uk/biomodels-main/termsofuse")->setDataset("http://identifiers.org/biomodels.db/");
$dataset_description .= $source_file->toRDF();
echo "done!" . PHP_EOL;
}
//foreach
parent::getWriteFile()->close();
$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}/")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/biomodels/biomodels.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 .= $output_file->toRDF();
//write dataset description to file
parent::setGraphURI($graph_uri);
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
}
function process_file($infile)
{
$odir = parent::getParameterValue('outdir');
$suffix = parent::getParameterValue('output_format');
$ofile = $odir . basename($infile, ".xml.gz") . '.' . $suffix;
$gz = strstr(parent::getParameterValue('output_format'), ".gz") === FALSE ? false : true;
$fp = gzopen($infile, "r") or die("Could not open file " . $infile . "!\n");
$this->setReadFile($infile);
$this->getReadFile()->setFilePointer($fp);
$this->setWriteFile($ofile, $gz);
$this->setCheckPoint('file');
$this->pubmed();
$this->writeRDFBufferToWriteFile();
$this->getWriteFile()->close();
}
function Run()
{
$sp = trim(parent::getParameterValue('files'));
if ($sp == 'all') {
$files = $this->getPackageMap();
} else {
$s_a = explode(",", $sp);
$pm = $this->getPackageMap();
$files = array();
foreach ($s_a as $a) {
if (array_key_exists($a, $pm)) {
$files[$a] = $pm[$a];
}
}
}
//else
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$dd = '';
//now iterate over the files array
$year = parent::getParameterValue('year');
foreach ($files as $k => $fpattern) {
$file = str_replace("YEAR", $year, $fpattern);
$lfile = $ldir . $file;
$rfile = parent::getParameterValue("download_url") . $file;
// download if necessary
if (!file_exists($lfile) || parent::getParameterValue('download') == "true") {
echo "Downloading {$file} ... ";
$ret = utils::downloadSingle($rfile, $lfile);
if ($ret === FALSE) {
trigger_error("Unable to get {$file}", E_USER_ERROR);
continue;
}
echo "done!" . PHP_EOL;
}
//set the outfile
$ofile = "mesh_" . $k . "." . parent::getParameterValue('output_format');
$gz = strstr(parent::getParameterValue('output_format'), "gz") ? true : false;
echo "processing {$k} ...";
parent::setReadFile($lfile, FALSE);
parent::setWriteFile($odir . $ofile, $gz);
$fnx = $k;
$this->{$fnx}();
parent::writeRDFBufferToWriteFile();
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("MeSH")->setRetrievedDate(parent::getDate(filemtime($lfile)))->setFormat("text/x-mesh-record")->setPublisher("http://www.nlm.nih.gov")->setHomepage("http://www.nlm.nih.gov/mesh/")->setRights("use")->setLicense("http://www.nlm.nih.gov/databases/download.html")->setDataset("http://identifiers.org/mesh/");
$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/mesh/mesh.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");
}
$dd .= $source_file->toRDF() . $output_file->toRDF();
}
//foreach
parent::setWriteFile($odir . $this->getBio2RDFReleaseFile($this->getNamespace()));
parent::getWriteFile()->write($dd);
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
}
function run()
{
if (parent::getParameterValue('files') == 'all') {
$files = explode("|", $this->GetParameterList('files'));
array_shift($files);
} else {
$files = explode(",", parent::getParameterValue('files'));
}
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$rdir = parent::getParameterValue('download_url');
$dataset_description = '';
foreach ($files as $file) {
$download = parent::getParameterValue('download');
$lfile = $ldir . "goa_" . $file . ".gz";
if (!file_exists($lfile) && $download == false) {
trigger_error($lfile . " not found. Will attempt to download.", E_USER_NOTICE);
$download = true;
}
//download file
$rfile = $rdir . strtoupper($file) . "/gene_association.goa_" . $file . ".gz";
if ($download == true) {
echo "downloading {$file} ... ";
//file_put_contents($lfile,file_get_contents($rfile));
utils::DownloadSingle($rfile, $lfile);
}
$gz = strstr(parent::getParameterValue('output_format'), ".gz") === FALSE ? false : true;
$ofile = "goa_" . $file . "." . parent::getParameterValue('output_format');
parent::setReadFile($lfile, TRUE);
parent::setWriteFile($odir . $ofile, $gz);
echo "processing {$file} ... ";
$this->process($file);
echo "done!";
parent::clear();
//close write file
parent::getWriteFile()->close();
echo PHP_EOL;
// dataset description
$graph_uri = parent::getGraphURI();
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("Gene Ontology Annotation file {$file} ({$rfile}")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat("text/tab-separated-value")->setFormat("application/gzip")->setPublisher("http://www.ebi.ac.uk/")->setHomepage("http://www.ebi.ac.uk/GOA/")->setRights("use")->setLicense("http://www.ebi.ac.uk/GOA/goaHelp.html")->setDataset("http://identifiers.org/goa/");
$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($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/irefindex/irefindex.php")->setCreateDate($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() . $output_file->toRDF();
parent::setGraphURI($graph_uri);
}
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
}
function getListOfEntries($ldir)
{
// get the master list of entries
$file = "mim2gene.txt";
if (!file_exists($ldir . $file)) {
trigger_error($ldir . $file . " not found. Will attempt to download. ", E_USER_NOTICE);
$this->SetParameterValue('download', true);
}
if (parent::getParameterValue('download') == true) {
// connect
if (!isset($ftp)) {
$host = 'ftp.omim.org';
echo "connecting to {$host} ...";
$ftp = ftp_connect($host);
if (!$ftp) {
echo "Unable to connect to {$host}" . PHP_EOL;
die;
}
ftp_pasv($ftp, true);
$login = ftp_login($ftp, 'anonymous', '*****@*****.**');
if (!$ftp || !$login) {
echo "FTP-connect failed!";
die;
} else {
echo "Connected" . PHP_EOL;
}
}
// download
ftp_pasv($ftp, true);
echo "Downloading {$file} ...";
if (ftp_get($ftp, $ldir . $file, 'OMIM/' . $file, FTP_BINARY) === FALSE) {
trigger_error("Error in downloading {$file}");
continue;
}
if (isset($ftp)) {
ftp_close($ftp);
}
echo "success!" . PHP_EOL;
}
// parse the mim2gene file for the entries
// # Mim Number Type Gene IDs Approved Gene Symbols
$fp = fopen($ldir . $file, "r");
fgets($fp);
while ($l = fgets($fp)) {
$a = explode("\t", $l);
if ($a[1] != "moved/removed") {
$list[$a[0]] = $a[1];
}
}
fclose($fp);
return $list;
}
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;
}
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 Run()
{
$dataset_description = '';
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$list_file = $ldir . "ftp_list.txt";
if (!file_exists($list_file) || $this->getParameterValue('download') == true) {
echo "Getting FTP file list ...";
$list = $this->getFtpFileList('ftp.ncbi.nlm.nih.gov', '/refseq/release/complete/', '/(complete\\.[0-9]+\\.protein\\.gpff\\.gz)/');
if (!isset($list) or count($list) == 0) {
trigger_error("Unable to get list of files from FTP site. Check internet connection", E_USER_ERROR);
exit(-1);
}
asort($list);
$buf = implode("\n", $list);
file_put_contents($list_file, $buf);
echo "Done." . PHP_EOL;
} else {
echo "Using existing ftp list" . PHP_EOL;
$list = explode("\n", file_get_contents($list_file));
}
$counter = 1;
$total = count($list);
foreach ($list as $f) {
$lfile = $ldir . $f;
echo "Processing " . $counter++ . "/{$total} {$f}. ";
if (!file_exists($lfile) || $this->getParameterValue('download') == true) {
$rfile = parent::getParameterValue('download_url') . $f;
echo "Downloading ...";
utils::DownloadSingle($rfile, $lfile);
echo "done.";
} else {
echo "Using existing file.";
}
echo PHP_EOL;
}
//if download
//iterate over the files
$files = $this->getFilePaths($ldir, 'gz');
asort($files);
foreach ($files as $f) {
$lfile = $ldir . $f;
$ofile = $odir . basename($f, ".gz") . "." . parent::getParameterValue('output_format');
$gz = strstr(parent::getParameterValue('output_format'), "gz") ? true : false;
parent::setWriteFile($ofile, $gz);
parent::setReadFile($lfile, true);
echo "processing {$f} ...";
$this->process();
parent::clear();
echo "done!" . PHP_EOL;
$this->getReadFile()->close();
$this->getWriteFile()->close();
$source_file = (new DataResource($this))->setURI(parent::getParameterValue('download_url') . $lfile)->setTitle("NCBI RefSeq - {$f}")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat('text/refseq-format')->setFormat('application/zip')->setPublisher('http://www.ncbi.nlm.nih.gov')->setHomepage('http://www.ncbi.nlm.nih.gov/refseq')->setRights('use')->setRights('attribution')->setLicense('http://www.nlm.nih.gov/copyright.html')->setDataset(parent::getDatasetURI());
$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}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} - {$f}")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/refseq/refseq.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() . $source_file->toRDF();
}
//for
parent::writeToReleaseFile($dataset_description);
parent::getWriteFile()->close();
}
function process()
{
if (parent::getParameterValue('files') == 'all') {
$files = explode("|", parent::getParameterList('files'));
array_shift($files);
} else {
$files = explode(",", parent::getParameterValue('files'));
}
$remote_files = array("human" => "human_genes.zip", "models" => "models_genes.zip");
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$rdir = parent::getParameterValue('download_url');
$dataset_description = '';
$graph_uri = parent::getGraphURI();
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
foreach ($files as $file) {
$lfile = $ldir . $remote_files[$file];
$rfile = $rdir . $remote_files[$file];
if (!file_exists($lfile)) {
trigger_error($lfile . " not found. Will attempt to download." . PHP_EOL, E_USER_WARNING);
echo "Downloading {$rfile}... ";
Utils::DownloadSingle($rfile, $lfile);
echo "done!" . PHP_EOL;
}
$suffix = parent::getParameterValue('output_format');
$ofile = "genage_" . $file . '.' . $suffix;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
$zin = new ZipArchive();
if ($zin->open($lfile) === FALSE) {
trigger_error("Unable to open {$lfile}");
exit;
}
if ($file == "human") {
$zipentry = "genage_human.csv";
} else {
if ($file == "models") {
$zipentry = "genage_models.csv";
}
}
if (($fp = $zin->getStream($zipentry)) === FALSE) {
trigger_error("Unable to get {$zipentry} in ziparchive {$lfile}");
return FALSE;
}
parent::SetReadFile($lfile);
parent::GetReadFile()->SetFilePointer($fp);
// set the write file, parse, write and close
$suffix = parent::getParameterValue('output_format');
$outfile = "genage_" . $file . '.' . $suffix;
$gz = false;
if (strstr($suffix, "gz")) {
$gz = true;
}
parent::setWriteFile($odir . $ofile, $gz);
echo "Processing {$lfile}... ";
$fnx = $file;
$this->{$fnx}();
echo "done!" . PHP_EOL;
parent::getWriteFile()->close();
// generate the dataset release file
echo "Generating dataset description for {$ofile}... ";
// dataset description
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("Human Ageing Genomic Resources GenAge database (" . $remote_files[$file] . ")")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat("text/comma-separated-value")->setFormat("application/gzip")->setPublisher("http://genomics.senescence.info/")->setHomepage("http://genomics.senescence.info/genes/")->setRights("use")->setLicense("http://genomics.senescence.info/legal.html")->setDataset("http://identifiers.org/genage/");
$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($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/genage/genage.php")->setCreateDate($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() . $output_file->toRDF();
echo "done!" . PHP_EOL;
}
parent::setGraphURI($graph_uri);
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
}
/**
* Function to begin parsing the local copy of the pubchem substances directory
**/
function parse_substances()
{
$ignore = array(".", "..");
$input_dir = $this->getParameterValue('indir') . "/substances/";
$gz = false;
$this->CreateDirectory($this->getParameterValue('outdir') . "/substances/");
parent::setDatasetURI("bio2rdf_dataset:bio2rdf-" . $this->getPcsPrefix() . "-" . date("Ymd"));
$graph_uri = parent::getGraphURI();
//set graph URI to dataset uri
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
$dataset_description = '';
if ($handle = opendir($input_dir)) {
while (false !== ($file = readdir($handle))) {
if (in_array($file, $ignore)) {
continue;
}
echo "Processing file: " . $input_dir . $file . PHP_EOL;
$suffix = parent::getParameterValue('output_format');
$outfile = realpath($this->getParameterValue('outdir')) . "/substances/" . basename($file, ".xml.gz") . "." . $suffix;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
echo "... into " . $outfile . PHP_EOL;
parent::setCheckpoint('file');
$this->setWriteFile($outfile, $gz);
$this->parse_substance_file($input_dir, $file);
$this->getWriteFile()->close();
}
closedir($handle);
$source_file = (new DataResource($this))->setURI("http://www.ncbi.nlm.nih.gov/pcsubstance")->setTitle("PubChem Substance")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($input_dir)))->setFormat("text/xml")->setFormat("application/zip")->setPublisher("http://ncbi.nlm.nih.gov/")->setHomepage("http://pubchem.ncbi.nlm.nih.gov/")->setRights("use")->setRights("restricted-by-source-license")->setLicense("ftp://ftp.ncbi.nlm.nih.gov/pubchem/README")->setDataset("http://identifiers.org/pubchem.substance/");
$prefix = $this->getPcsPrefix();
$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}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/pubchem/pubchem.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() . $output_file->toRDF();
//set graph URI back to default
parent::setGraphURI($graph_uri);
// write the dataset description
$this->setWriteFile($this->getParameterValue('outdir') . "/substances/" . $this->getBio2RDFReleaseFile());
$this->getWriteFile()->write($dataset_description);
$this->getWriteFile()->close();
} else {
echo "unable to read directory contents: " . $input_dir . "\n";
exit;
}
}
function Run()
{
// directory shortcuts
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
// get the listings page
$url = trim(parent::getParameterValue('download_url'));
$listing_file = $ldir . "probeset_list.html";
if (!file_exists($listing_file) || parent::getParameterValue("download") == "true") {
echo "Downloading {$listing_file}" . PHP_EOL;
Utils::DownloadSingle($url, $listing_file);
}
$listings = file_get_contents($listing_file);
// make a list of the csv.zip files
preg_match_all("/\"([^\"]+)\\.csv\\.zip\"/", $listings, $m);
if (count($m[1]) == 0) {
trigger_error("could not find any .csv.zip files in {$url}");
exit;
}
if (parent::getParameterValue("files") == 'all') {
$myfiles = $m[1];
} else {
$a = explode(",", parent::getParameterValue("files"));
foreach ($a as $f) {
$found = false;
foreach ($m[1] as $n) {
if (strstr($n, $f)) {
$found = true;
$myfiles[] = $n;
break;
}
}
if ($found === false) {
echo "cannot find {$f} in list" . PHP_EOL;
}
}
}
if (!isset($myfiles)) {
exit;
}
// nothing to do
$dataset_description = '';
// set the write file
$gz = strstr(parent::getParameterValue('output_format'), ".gz") === FALSE ? false : true;
$outfile = 'affymetrix.' . parent::getParameterValue('output_format');
$this->setWriteFile($odir . $outfile, $gz);
// iterate over the files
foreach ($myfiles as $rfile) {
$base_file = substr($rfile, strrpos($rfile, "/") + 1);
$base_url = substr($rfile, 0, strrpos($rfile, "/"));
// get and set the dataset version
if (parent::getDatasetVersion() == null) {
preg_match("/\\.na([0-9]{2})\\.annot/", $base_file, $m);
if (isset($m[1])) {
$this->setDatasetVersion($m[1]);
}
}
if (parent::getDatasetVersion() != parent::getParameterValue('version')) {
$base_file = str_replace("na" . parent::getDatasetVersion(), "na" . parent::getParameterValue('version'), $base_file);
}
$csv_file = $base_file . ".csv";
$zip_file = $csv_file . ".zip";
$lfile = $ldir . $zip_file;
if (!file_exists($lfile)) {
echo "skipping: {$lfile} does not exist" . PHP_EOL;
continue;
}
echo "processing {$lfile}" . PHP_EOL;
// open the zip file
$zin = new ZipArchive();
if ($zin->open($lfile) === FALSE) {
trigger_error("Unable to open {$lfile}");
exit;
}
if (($fp = $zin->getStream($csv_file)) === FALSE) {
trigger_error("Unable to get {$csv_file} in ziparchive {$lfile}");
return FALSE;
}
parent::setReadFile($lfile);
parent::getReadFile()->setFilePointer($fp);
$this->parse($base_file);
parent::getReadFile()->close();
parent::clear();
// dataset description
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("Affymetrix Probeset: {$base_file}")->setRetrievedDate(parent::getDate(filemtime($lfile)))->setFormat("text/tab-separated-value")->setFormat("application/zip")->setPublisher("http://affymetrix.com")->setHomepage("http://www.affymetrix.com/support/technical/annotationfilesmain.affx")->setRights("use")->setRights("no-commercial")->setRights("registration-required")->setLicense("http://www.affymetrix.com/about_affymetrix/legal/index.affx")->setDataset("http://identifiers.org/affy.probeset/");
$dataset_description .= $source_file->toRDF();
}
$this->getWriteFile()->close();
// write the dataset description
$prefix = parent::getPrefix();
$bVersion = parent::getParameterValue('bio2rdf_release');
$date = parent::getDate(filemtime($odir . $outfile));
$output_file = (new DataResource($this))->setURI("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$outfile}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix}")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/affymetrix/affymetrix.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 .= $output_file->toRDF();
// write the dataset description
$this->setWriteFile($odir . $this->getBio2RDFReleaseFile());
$this->getWriteFile()->write($dataset_description);
$this->getWriteFile()->close();
return true;
}
function process()
{
// get the file list
if (parent::getParameterValue('files') == 'all') {
$files = explode("|", parent::getParameterList('files'));
array_shift($files);
} else {
$files = explode(",", parent::getParameterValue('files'));
}
$dataset_description = '';
//set directory values
$ldir = parent::getParameterValue('indir');
$rdir = parent::getParameterValue('download_url');
$odir = parent::getParameterValue('outdir');
$graph_uri = parent::getGraphURI();
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
$gz_suffix = ".gz";
foreach ($files as $file) {
if ($file == 'chem_gene_ixn_types') {
$suffix = '.tsv';
} else {
if ($file == 'exposure_ontology') {
$suffix = '.obo';
} else {
$suffix = ".tsv.gz";
}
}
$lfile = $ldir . $file . $gz_suffix;
$rfile = $rdir . 'CTD_' . $file . $suffix;
if (!file_exists($lfile)) {
trigger_error($lfile . " not found. Will attempt to download.", E_USER_NOTICE);
if ($suffix == ".tsv.gz") {
Utils::DownloadSingle($rfile, $lfile);
} else {
Utils::DownloadSingle($rfile, "compress.zlib://" . $lfile);
}
}
$out_suffix = parent::getParameterValue('output_format');
$ofile = "ctd_" . $file . "." . $out_suffix;
$gz = false;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
echo "Processing " . $file . " ...";
parent::setWriteFile($odir . $ofile, $gz);
//set read file
parent::setReadFile($lfile, TRUE);
$fnx = "CTD_" . $file;
$this->{$fnx}();
//close write file
parent::getWriteFile()->close();
parent::clear();
echo "done!" . PHP_EOL;
// generate the dataset release file
echo "Generating dataset description... ";
if ($file == "chemicals") {
$dataset = "http://identifiers.org/ctd.chemical/";
} else {
if ($file == "diseases") {
$dataset = "http://identifiers.org/ctd.disease/";
} else {
if ($file == "genes") {
$dataset = "http://identifiers.org/ctd.gene/";
} else {
$dataset = null;
}
}
}
// dataset description
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("Comparative Toxicogenomics Database ({$file}.{$gz_suffix}")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat("text/tab-separated-value")->setFormat("application/gzip")->setPublisher("http://ctdbase.org/")->setHomepage("http://ctdbase.org/")->setRights("use")->setRights("by-attribution")->setRights("no-commercial")->setLicense("http://ctdbase.org/about/legal.jsp")->setDataset($dataset);
$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($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/ctd/ctd.php")->setCreateDate($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() . $output_file->toRDF();
}
parent::setGraphURI($graph_uri);
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
}
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($xml)
{
// state the dataset info
foreach ($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})"));
}
}
}
function Run()
{
// get the work
if ($this->GetParameterValue('files') == 'all') {
$sources = explode("|", parent::getParameterList('files'));
array_shift($sources);
} else {
// comma separated list
$sources = explode(",", parent::getParameterValue('files'));
}
$download_files = array("h**o-sapiens" => "Pathway%20Commons%202%20homo%20sapiens.BIOPAX.owl.gz", "hprd" => "Pathway%20Commons%202%20HPRD.BIOPAX.owl.gz", "humancyc" => "Pathway%20Commons%202%20HumanCyc.BIOPAX.owl.gz", "nci-nature" => "Pathway%20Commons%202%20NCI_Nature.BIOPAX.owl.gz", "panther-pathway" => "Pathway%20Commons%202%20PANTHER%20Pathway.BIOPAX.owl.gz", "phosphositeplus" => "Pathway%20Commons%202%20PhosphoSitePlus.BIOPAX.owl.gz", "reactome" => "Pathway%20Commons%202%20Reactome.BIOPAX.owl.gz");
$graph_uri = parent::getGraphURI();
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
$dataset_description = '';
// iterate over the requested data
foreach ($sources as $source) {
echo "processing {$source}... ";
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$rdir = parent::getParameterValue('download_url');
// set the remote and input files
$file = $source . ".owl";
$zfile = $source . ".owl.gz";
$rfile = $rdir . $download_files[$source];
$lfile = $ldir . $zfile;
// download if if the file doesn't exist locally or we are told to
if (!file_exists($lfile) || $this->GetParameterValue('download') == 'true') {
// download
echo "downloading... ";
file_put_contents($lfile, file_get_contents($rfile));
}
// extract the file out of the ziparchive
// and load into a buffer
echo 'extracting... ';
if (($fpin = gzopen($lfile, "r")) === FALSE) {
trigger_error("Unable to open {$lfile}", E_USER_ERROR);
exit;
}
$data = '';
while (!gzeof($fpin)) {
$buffer = gzgets($fpin, 4096);
$data .= $buffer;
}
gzclose($fpin);
// set the output file
$suffix = parent::getParameterValue('output_format');
$outfile = $source . '.' . $suffix;
$gz = false;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
parent::setWriteFile($odir . $outfile, $gz);
// send for parsing
$p = new BioPAX2Bio2RDF($this);
$p->SetBuffer($data)->SetBioPAXVersion(3)->SetBaseNamespace("http://purl.org/pc2/3/")->SetBio2RDFNamespace("http://bio2rdf.org/pathwaycommons:")->SetDatasetURI(parent::getDatasetURI());
$rdf = $p->Parse();
parent::addRDF($rdf);
// write to output
parent::writeRDFBufferToWriteFile();
parent::getWriteFile()->Close();
echo "done!" . PHP_EOL;
//generate dataset description
echo "Generating dataset description for {$zfile}... ";
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("Pathway Commons")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat("rdf/xml")->setPublisher("http://www.pathwaycommons.org/")->setHomepage("http://www.pathwaycommons.org/")->setRights("use")->setRights("restricted-by-source-license")->setLicense("http://www.pathwaycommons.org/pc2/home.html#data_sources")->setDataset("http://identifiers.org/pathwaycommons/");
$dataset_description .= $source_file->toRDF();
echo "done!" . PHP_EOL;
}
echo "Generating dataset description for Bio2RDF Pathways Commons dataset... ";
$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}/")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/pathwaycommons/pathwaycommons.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 .= $output_file->toRDF();
//write dataset description to file
parent::setGraphURI($graph_uri);
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
}
