function process_dir()
{
$this->setCheckPoint('dataset');
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$this->id_list = null;
if (parent::getParameterValue('id_list') != '') {
$this->id_list = array_flip(explode(",", trim(parent::getParameterValue("id_list"))));
}
$graph_uri = parent::getGraphURI();
$dataset_description = '';
$gz = strstr(parent::getParameterValue('output_format'), ".gz") === FALSE ? false : true;
//set graph URI to dataset graph
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
$files = glob($ldir . "*.xml.gz");
foreach ($files as $i => $file) {
echo "Processing {$file} (" . ($i + 1) . "/" . count($files) . ") ...";
$this->process_file($file);
parent::clear();
echo "done!" . PHP_EOL;
}
$source_file = (new DataResource($this))->setURI("http://www.ncbi.nlm.nih.gov/pubmed")->setTitle("NCBI PubMed")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($ldir)))->setFormat("text/xml")->setPublisher("http://ncbi.nlm.nih.gov/")->setHomepage("http://www.ncbi.nlm.nih.gov/pubmed/")->setRights("use-share-modify")->setLicense("http://www.nlm.nih.gov/databases/license/license.html")->setDataset("http://identifiers.org/pubmed/");
$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/pubmed/pubmed.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
$this->setWriteFile($odir . $this->getBio2RDFReleaseFile());
$this->getWriteFile()->write($dataset_description);
$this->getWriteFile()->close();
}
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;
}
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 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 process()
{
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
$rdir = parent::getParameterValue('download_url');
//which files are to be converted?
$files = trim($this->GetParameterValue('files'));
if ($files == 'all') {
$files = $this->getPackageMap();
} else {
$sel_arr = explode(",", $files);
$pm = $this->getPackageMap();
$files = array();
foreach ($sel_arr as $a) {
if (array_key_exists($a, $pm)) {
$files[$a] = $pm[$a];
}
}
}
if ($this->getParameterValue('limit_organisms') == true) {
$this->taxids = array_flip(explode(",", $this->getParameterValue('organisms')));
}
//set dataset graph to be dataset URI
$graph_uri = parent::getGraphURI();
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
$dataset_description = '';
//now iterate over the files array
foreach ($files as $module => $rfilename) {
$file = $module . ".gz";
$lfile = $ldir . $file;
$rfile = $rdir . $rfilename;
// download
if (!file_exists($lfile) || parent::getParameterValue('download') == true) {
trigger_error("{$lfile} not found. Will attempt to download.", E_USER_NOTICE);
$myfile = $lfile;
if ($module == "gene2sts" || $module == "gene2unigene") {
$myfile = "compress.zlib://" . $lfile;
}
echo "downloading {$module} ...";
utils::DownloadSingle($rfile, $myfile);
echo "done" . PHP_EOL;
}
}
foreach ($files as $module => $rfilename) {
$file = $module . ".gz";
$lfile = $ldir . $file;
$rfile = $rdir . $rfilename;
$ofile = $module . "." . parent::getParameterValue('output_format');
$gz = false;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
echo "Processing {$module} ... ";
parent::setReadFile($lfile, true);
parent::setWriteFile($odir . $ofile, $gz);
$fnx = $module;
if ($module == 'gene2refseq') {
$fnx = 'gene2accession';
}
$this->{$fnx}();
parent::clear();
echo 'done!' . PHP_EOL;
parent::getReadFile()->close();
parent::getWriteFile()->close();
// generate the dataset release file
// dataset description
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("NCBI Gene ({$module})")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat("text/tab-separated-value")->setFormat("application/gzip")->setPublisher("http://www.ncbi.nlm.nih.gov")->setHomepage("http://www.ncbi.nlm.nih.gov/gene")->setRights("use-share-modify")->setLicense("http://www.ncbi.nlm.nih.gov/About/disclaimer.html")->setDataset("http://identifiers.org/ncbigene/");
$prefix = parent::getPrefix();
$bVersion = parent::getParameterValue('bio2rdf_release');
$date = date("Y-m-d\\TG:i:s\\Z");
$output_file = (new DataResource($this))->setURI("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$ofile}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/ncbigene/ncbigene.php")->setCreateDate($date)->setHomepage("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$prefix}.html")->setPublisher("http://bio2rdf.org")->setRights("use-share-modify")->setRights("by-attribution")->setRights("restricted-by-source-license")->setLicense("http://creativecommons.org/licenses/by/3.0/")->setDataset(parent::getDatasetURI());
if ($gz) {
$output_file->setFormat("application/gzip");
}
if (strstr(parent::getParameterValue('output_format'), "nt")) {
$output_file->setFormat("application/n-triples");
} else {
$output_file->setFormat("application/n-quads");
}
$dataset_description .= $source_file->toRDF() . $output_file->toRDF();
}
//foreach
//set graph URI back to default value
parent::setGraphURI($graph_uri);
//write dataset description to file
echo "Generating dataset description... ";
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
}
function 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 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 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;
}
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 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 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;
}
function process()
{
if (parent::getParameterValue('files') == 'all') {
$files = explode("|", parent::getParameterList('files'));
array_shift($files);
} else {
$files = explode(",", parent::getParameterValue('files'));
}
$ldir = parent::getParameterValue('indir');
$rdir = parent::getParameterValue('download_url');
$odir = parent::getParameterValue('outdir');
$rfiles = array("dbxref" => "curation/chromosomal_feature/dbxref.tab", "features" => "curation/chromosomal_feature/SGD_features.tab", "domains" => "curation/calculated_protein_info/domains/domains.tab", "protein" => "curation/calculated_protein_info/protein_properties.tab", "goa" => "curation/literature/gene_association.sgd.gz", "goslim" => "curation/literature/go_slim_mapping.tab", "complex" => "curation/literature/go_protein_complex_slim.tab", "interaction" => "curation/literature/interaction_data.tab", "phenotype" => "curation/literature/phenotype_data.tab", "pathways" => "curation/literature/biochemical_pathways.tab", "mapping" => "mapping");
$graph_uri = parent::getGraphURI();
if (parent::getParameterValue('dataset_graph') == true) {
parent::setGraphURI(parent::getDatasetURI());
}
$gz = false;
if (strstr(parent::getParameterValue('output_format'), "gz")) {
$gz = true;
}
if (parent::getParameterValue('one_file') == true) {
$ofile = "sgd." . parent::getParameterValue('output_format');
parent::setWriteFile($odir . $ofile, $gz);
}
$dataset_description = '';
foreach ($files as $file) {
$ext = substr(strrchr($rfiles[$file], '.'), 1);
if ($ext == "tab") {
$lfile = "sgd_" . $file . ".tab";
} elseif ($ext = "gz") {
$lfile = "sgd_" . $file . ".tab.gz";
}
$rfile = $rdir . $rfiles[$file];
if (!file_exists($ldir . $lfile) && parent::getParameterValue('download') == false && $file != 'mapping') {
trigger_error($ldir . $lfile . " not found. Will attempt to download.", E_USER_NOTICE);
Utils::DownloadSingle($rfile, $ldir . $lfile);
}
if (parent::getParameterValue('one_file') == false) {
$ofile = "sgd_" . $file . '.' . parent::getParameterValue('output_format');
parent::setWriteFile($odir . $ofile, $gz);
}
//parse file
parent::setReadFile($ldir . $lfile, $gz);
$fnx = $file;
echo "Processing {$file}... ";
$this->{$fnx}();
echo PHP_EOL . "done!";
//write RDF to file
parent::writeRDFBufferToWriteFile();
//close write file
if (parent::getParameterValue('one_file') == false) {
parent::getWriteFile()->close();
}
echo PHP_EOL;
// generate the dataset release file
// dataset description
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("Saccharomyces Genome Database ({$file})")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($ldir . $lfile)))->setFormat("text/tab-separated-value")->setFormat("application/gzip")->setPublisher("http://www.yeastgenome.org/")->setHomepage("http://www.yeastgenome.org/")->setRights("use")->setLicense("http://www.stanford.edu/site/terms.html")->setDataset("http://identifiers.org/sgd/");
$dataset_description .= $source_file->toRDF();
if (parent::getParameterValue('one_file') == false) {
$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($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/sgd/sgd.php")->setCreateDate($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 .= $output_file->toRDF();
}
}
//foreach
//set graph URI back to default
parent::setGraphURI($graph_uri);
if (parent::getParameterValue('one_file') == true) {
$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($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/sgd/sgd.php")->setCreateDate($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() . $output_file->toRDF();
}
//write dataset description to file
echo "Generating dataset description... " . PHP_EOL;
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
echo "done!" . PHP_EOL;
}
function Run()
{
$indir = parent::getParameterValue('indir');
$outdir = parent::getParameterValue('outdir');
$download_url = parent::getParameterValue('download_url');
if (parent::getParameterValue('files') == 'all') {
$files = explode("|", parent::getParameterList('files'));
array_shift($files);
} else {
$files = explode("|", parent::getParameterValue('files'));
}
if (parent::getParameterValue("id_list")) {
$this->id_list = array_flip(explode(",", parent::getParameterValue('id_list')));
}
$dataset_description = '';
foreach ($files as $f) {
if ($f == 'drugbank') {
$file = 'drugbank.xml.zip';
$lname = 'drugbank';
}
$fnx = 'parse_' . $f;
$rfile = parent::getParameterValue('download_url') . $file;
$lfile = parent::getParameterValue('indir') . $file;
$cfile = $lname . "." . parent::getParameterValue('output_format');
// download
if (!file_exists($lfile) || parent::getParameterValue('download') == true) {
utils::downloadSingle($rfile, $lfile);
}
// setup the write
$gz = strstr(parent::getParameterValue('output_format'), ".gz") === FALSE ? false : true;
parent::setWriteFile($outdir . $cfile, $gz);
echo $outdir . $cfile;
if (file_exists($indir . $file)) {
// call the parser
echo "processing {$file} ..." . PHP_EOL;
$this->{$fnx}($indir, $file);
echo "done" . PHP_EOL;
parent::clear();
}
parent::getWriteFile()->close();
// dataset description
$ouri = parent::getGraphURI();
parent::setGraphURI(parent::getDatasetURI());
$source_version = parent::getDatasetVersion();
$bVersion = parent::getParameterValue('bio2rdf_release');
$prefix = parent::getPrefix();
$date = date("Y-m-d\\TH:i:sP");
// dataset description
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("DrugBank ({$file})")->setRetrievedDate(date("Y-m-d\\TH:i:sP", filemtime($indir . $file)))->setFormat("application/xml")->setFormat("application/zip")->setPublisher("http://drugbank.ca")->setHomepage("http://drugbank.ca")->setRights("use")->setRights("by-attribution")->setRights("no-commercial")->setLicense("http://www.drugbank.ca/about")->setDataset("http://identifiers.org/drugbank/");
$output_file = (new DataResource($this))->setURI("http://download.bio2rdf.org/release/{$bVersion}/{$prefix}/{$cfile}")->setTitle("Bio2RDF v{$bVersion} RDF version of {$prefix} v{$source_version}")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/drugbank/drugbank.php")->setCreateDate($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 {
$output_file->setFormat("application/n-quads");
}
parent::writeToReleaseFile($source_file->toRDF() . $output_file->toRDF());
parent::setGraphURI($ouri);
}
parent::closeReleaseFile();
}
function OBO2RDF($abbv)
{
$abbv = strtolower($abbv);
if ($abbv == "doid") {
$abbv = "do";
}
$minimal = parent::getParameterValue('detail') == 'min' ? true : false;
$minimalp = parent::getParameterValue('detail') == 'min+' ? true : false;
$version = parent::getParameterValue("bio2rdf_release");
$tid = '';
$first = true;
$is_a = false;
$is_deprecated = false;
$min = $buf = '';
$ouri = "http://bio2rdf.org/lsr:" . $abbv;
$dataset_uri = $abbv . "_resource:bio2rdf.dataset.{$abbv}.R" . $version;
parent::setGraphURI($dataset_uri);
$buf = parent::triplify($ouri, "rdf:type", "owl:Ontology");
$graph_uri = '<' . parent::getRegistry()->getFQURI(parent::getGraphURI()) . '>';
$bid = 1;
while ($l = parent::getReadFile()->read()) {
$lt = trim($l);
if (strlen($lt) == 0) {
continue;
}
if ($lt[0] == '!') {
continue;
}
if (strstr($l, "[Term]")) {
// first node?
if ($first == true) {
// ignore the first case
$first = false;
} else {
if ($tid != '' && $is_a == false && $is_deprecated == false) {
$t = parent::triplify($tid, "rdfs:subClassOf", "obo_vocabulary:Entity");
$buf .= $t;
$min .= $t;
}
}
$is_a = false;
$is_deprecated = false;
unset($typedef);
$term = '';
$tid = '';
continue;
} else {
if (strstr($l, "[Typedef]")) {
$is_a = false;
$is_deprecated = false;
unset($term);
$tid = '';
$typedef = '';
continue;
}
}
//echo "LINE: $l".PHP_EOL;
// to fix error in obo generator
$lt = str_replace("synonym ", "synonym: ", $lt);
$lt = preg_replace("/\\{.*\\} !/", " !", $lt);
$a = explode(" !", $lt);
if (isset($a[1])) {
$exc = trim($a[1]);
}
$a = explode(": ", trim($a[0]), 2);
// let's go
if (isset($intersection_of)) {
if ($a[0] != "intersection_of") {
// $intersection_of .= ")].".PHP_EOL;
//$buf .= $intersection_of;
if ($minimalp) {
$min .= $intersection_of;
}
unset($intersection_of);
}
}
if (isset($relationship)) {
if ($a[0] != "relationship") {
// $relationship .= ")].".PHP_EOL;
//$buf .= $relationship;
if ($minimalp) {
$min .= $relationship;
}
unset($relationship);
}
}
if (isset($typedef)) {
if ($a[0] == "id") {
$c = explode(":", $a[1]);
if (count($c) == 1) {
$ns = "obo";
$id = $c[0];
} else {
$ns = strtolower($c[0]);
$id = $c[1];
}
$id = str_replace(array("(", ")"), array("_", ""), $id);
$tid = $ns . ":" . $id;
} else {
if ($a[0] == "name") {
$buf .= parent::describeClass($tid, addslashes(stripslashes($a[1])));
} else {
if ($a[0] == "is_a") {
if (FALSE !== ($pos = strpos($a[1], "!"))) {
$a[1] = substr($a[1], 0, $pos - 1);
}
$buf .= parent::triplify($tid, "rdfs:subPropertyOf", "obo_vocabulary:" . strtolower($a[1]));
} else {
if ($a[0] == "is_obsolete") {
$buf .= parent::triplify($tid, "rdf:type", "owl:DeprecatedClass");
$is_deprecated = true;
} else {
if ($a[0][0] == "!") {
$a[0] = substr($a[0], 1);
}
$buf .= parent::triplifyString($tid, "obo_vocabulary:{$a['0']}", str_replace('"', '', stripslashes($a[1])));
}
}
}
}
} else {
if (isset($term)) {
if ($a[0] == "is_obsolete" && $a[1] == "true") {
$t = parent::triplify($tid, "rdf:type", "owl:DeprecatedClass");
$t .= parent::triplify($tid, "rdfs:subClassOf", "owl:DeprecatedClass");
$min .= $t;
$buf .= $t;
$is_deprecated = true;
} else {
if ($a[0] == "id") {
parent::getRegistry()->parseQName($a[1], $ns, $id);
$tid = "{$ns}:{$id}";
// $buf .= parent::describeClass($tid,null,"owl:Class");
// $buf .= parent::triplify($tid,"rdfs:isDefinedBy",$ouri);
} else {
if ($a[0] == "name") {
// $t = parent::triplifyString($tid,"rdfs:label",str_replace(array("\"", "'"), array("","\\\'"), stripslashes($a[1]))." [$tid]");
$label = str_replace(array("\"", "'"), array("", "\\\\'"), stripslashes($a[1]));
$t = parent::describeIndividual($tid, $label, "owl:Class");
$t .= parent::triplify($tid, "rdfs:isDefinedBy", $ouri);
$min .= $t;
$buf .= $t;
} else {
if ($a[0] == "def") {
$t = str_replace(array("'", "\"", "\\", "\\\\'"), array("\\\\'", "", "", ""), $a[1]);
$min .= parent::triplifyString($tid, "dc:description", $t);
$buf .= parent::triplifyString($tid, "dc:description", $t);
} else {
if ($a[0] == "property_value") {
$b = explode(" ", $a[1]);
$buf .= parent::triplifyString($tid, "obo_vocabulary:" . strtolower($b[0]), str_replace("\"", "", strtolower($b[1])));
} else {
if ($a[0] == "xref") {
// http://upload.wikimedia.org/wikipedia/commons/3/34/Anatomical_Directions_and_Axes.JPG
// Medical Dictionary:http\://www.medterms.com/
// KEGG COMPOUND:C02788 "KEGG COMPOUND"
// id-validation-regexp:\"REACT_[0-9\]\{1\,4}\\.[0-9\]\{1\,3}|[0-9\]+\"
//$a[1] = 'id-validation-regexp:\"REACT_[0-9\]\{1\,4}\\.[0-9\]\{1\,3}|[0-9\]+\"';
if (substr($a[1], 0, 4) == "http") {
$buf .= parent::triplify($tid, "rdfs:seeAlso", str_replace(array(" ", '"wiki"', "\\"), array("+", "", ""), $a[1]));
} else {
$b = explode(":", $a[1], 2);
if (substr($b[1], 0, 4) == "http") {
$buf .= parent::triplify($tid, "rdfs:seeAlso", stripslashes($b[1]));
} else {
$ns = str_replace(array(" ", "\\"), "", strtolower($b[0]));
$id = trim($b[1]);
// there may be a comment to remove
if (FALSE !== ($pos = strrpos($id, ' "'))) {
$comment = substr($id, $pos + 1, -1);
$id = substr($id, 0, $pos);
}
$id = stripslashes($id);
// there may be a source statement to remove
$id = preg_replace("/{.*\\}/", "", $id);
if ($ns == "pmid") {
$ns = "pubmed";
$y = explode(" ", $id);
$id = $y[0];
}
if ($ns == "xx") {
continue;
}
if ($ns == "icd9cm") {
$y = explode(" ", $id);
$id = $y[0];
}
if ($ns == "xref; umls_cui") {
continue;
}
if ($ns == "submitter") {
$ns = "chebi.submitter";
}
if ($ns == "wikipedia" || $ns == "mesh") {
$id = str_replace(" ", "+", $id);
}
if ($ns == "id-validation-regexp") {
$buf .= parent::triplifyString($tid, "obo_vocabulary:{$ns}", addslashes($id));
} else {
$buf .= parent::triplify($tid, "obo_vocabulary:x-{$ns}", "{$ns}:" . str_replace(" ", "-", $id));
}
}
}
} else {
if ($a[0] == "synonym") {
// synonym: "entidades moleculares" RELATED [IUPAC:]
// synonym: "molecular entity" EXACT IUPAC_NAME [IUPAC:]
// synonym: "Chondrococcus macrosporus" RELATED synonym [NCBITaxonRef:Krzemieniewska_and_Krzemieniewski_1926]
//grab string inside double quotes
preg_match('/"(.*)"(.*)/', $a[1], $matches);
if (!empty($matches)) {
$a[1] = str_replace(array("\\", "\"", "'"), array("", "", "\\\\'"), $matches[1] . $matches[2]);
} else {
$a[1] = str_replace(array("\"", "'"), array("", "\\\\'"), $a[1]);
}
$rel = "SYNONYM";
$list = array("EXACT", "BROAD", "RELATED", "NARROW");
$found = false;
foreach ($list as $keyword) {
// get everything after the keyword up until the bracket [
if (FALSE !== ($k_pos = strpos($a[1], $keyword))) {
$str_len = strlen($a[1]);
$keyword_len = strlen($keyword);
$keyword_end_pos = $k_pos + $keyword_len;
$b1_pos = strrpos($a[1], "[");
$b2_pos = strrpos($a[1], "]");
$b_text = substr($a[1], $b1_pos + 1, $b2_pos - $b1_pos - 1);
$diff = $b1_pos - $keyword_end_pos - 1;
if ($diff != 0) {
// then there is more stuff here
$k = substr($a[1], $keyword_end_pos + 1, $diff);
$rel = trim($k);
} else {
// create the long predicate
$rel = $keyword . "_SYNONYM";
}
$found = true;
$str = substr($a[1], 0, $k_pos - 1);
break;
}
}
// check to see if we still haven't found anything
if ($found === false) {
// we didn't find one of the keywords
// so take from the start to the bracket
$b1_pos = strrpos($a[1], "[");
$str = substr($a[1], 0, $b1_pos - 1);
}
$rel = str_replace(" ", "_", $rel);
// $lit = addslashes($str.($b_text?" [".$b_text."]":""));
$l = parent::triplifyString($tid, "obo_vocabulary:" . strtolower($rel), $str);
$buf .= $l;
} else {
if ($a[0] == "alt_id") {
parent::getRegistry()->parseQname($a[1], $ns, $id);
if ($id != 'curators') {
$buf .= parent::triplify("{$ns}:{$id}", "rdfs:seeAlso", stripslashes($tid));
}
} else {
if ($a[0] == "is_a") {
// do subclassing
parent::getRegistry()->parseQName($a[1], $ns, $id);
$t = parent::triplify($tid, "rdfs:subClassOf", "{$ns}:{$id}");
$buf .= $t;
$min .= $t;
$is_a = true;
} else {
if ($a[0] == "intersection_of") {
if (!isset($intersection_of)) {
// $intersection_of = '<'.parent::getRegistry()->getFQURI($tid).'> <'.parent::getRegistry()->getFQURI('owl:equivalentClass').'> [<'.parent::getRegistry()->getFQURI('rdf:type').'> <'.parent::getRegistry()->getFQURI('owl:Class').'>; <'.parent::getRegistry()->getFQURI('owl:intersectionOf').'> (';
$intersection_of = '<' . parent::getRegistry()->getFQURI($tid) . '> <' . parent::getRegistry()->getFQURI('owl:equivalentClass') . '> _:b' . ++$bid . " {$graph_uri} ." . PHP_EOL;
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('rdf:type') . '> <' . parent::getRegistry()->getFQURI('owl:Class') . "> {$graph_uri} ." . PHP_EOL;
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:intersectionOf') . '> _:b' . ++$bid . " {$graph_uri} ." . PHP_EOL;
}
/*
intersection_of: ECO:0000206 ! BLAST evidence
intersection_of: develops_from VAO:0000092 ! chondrogenic condensation
intersection_of: OBO_REL:has_part VAO:0000040 ! cartilage tissue
*/
$c = explode(" ", $a[1]);
if (count($c) == 1) {
// just a class
parent::getRegistry()->parseQName($c[0], $ns, $id);
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('rdfs:subClassOf') . '> <' . parent::getRegistry()->getFQURI("{$ns}:{$id}") . "> {$graph_uri} ." . PHP_EOL;
$buf .= parent::triplify($tid, "rdfs:subClassOf", "{$ns}:{$id}");
} else {
if (count($c) == 2) {
// an expression
parent::getRegistry()->parseQName($c[0], $pred_ns, $pred_id);
parent::getRegistry()->parseQName($c[1], $obj_ns, $obj_id);
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:onProperty') . '> <' . parent::getRegistry()->getFQURI("obo_vocabulary:" . $pred_id) . "> {$graph_uri} ." . PHP_EOL;
$intersection_of .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:someValuesFrom') . '> <' . parent::getRegistry()->getFQURI("{$obj_ns}:{$obj_id}") . "> {$graph_uri} ." . PHP_EOL;
$buf .= parent::triplify($tid, "obo_vocabulary:{$pred_id}", "{$obj_ns}:{$obj_id}");
}
}
} else {
if ($a[0] == "relationship") {
if (!isset($relationship)) {
$relationship = '<' . parent::getRegistry()->getFQURI($tid) . '> <' . parent::getRegistry()->getFQURI('rdfs:subClassOf') . '> _:b' . ++$bid . " {$graph_uri} ." . PHP_EOL;
$relationship .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('rdf:type') . '> <' . parent::getRegistry()->getFQURI('owl:Class') . "> {$graph_uri} ." . PHP_EOL;
$relationship .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:intersectionOf') . '> _:b' . ++$bid . " {$graph_uri} ." . PHP_EOL;
}
/*
relationship: develops_from VAO:0000092 ! chondrogenic condensation
relationship: OBO_REL:has_part VAO:0000040 ! cartilage tissue
*/
$c = explode(" ", $a[1]);
if (count($c) == 1) {
// just a class
parent::getRegistry()->parseQName($c[0], $ns, $id);
$relationship .= parent::getRegistry()->getFQURI("{$ns}:{$id}");
$buf .= parent::triplify($tid, "rdfs:subClassOf", "{$ns}:{$id}");
} else {
if (count($c) == 2) {
// an expression
parent::getRegistry()->parseQName($c[0], $pred_ns, $pred_id);
parent::getRegistry()->parseQName($c[1], $obj_ns, $obj_id);
$relationship .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:onProperty') . '> <' . parent::getRegistry()->getFQURI("obo_vocabulary:" . $pred_id) . "> {$graph_uri} ." . PHP_EOL;
$relationship .= '_:b' . $bid . ' <' . parent::getRegistry()->getFQURI('owl:someValuesFrom') . '> <' . parent::getRegistry()->getFQURI("{$obj_ns}:{$obj_id}") . "> {$graph_uri} ." . PHP_EOL;
$buf .= parent::triplify($tid, "obo_vocabulary:{$pred_id}", "{$obj_ns}:{$obj_id}");
}
}
} else {
// default handler
if (isset($a[1])) {
$buf .= parent::triplifyString($tid, "obo_vocabulary:{$a['0']}", str_replace(array("\"", "'"), array("", "\\\\'"), stripslashes($a[1])));
}
}
}
}
}
}
}
}
}
}
}
}
} else {
//header
//format-version: 1.0
$buf .= parent::triplifyString($ouri, "obo_vocabulary:{$a['0']}", str_replace(array('"', '\\:'), array('\\"', ':'), isset($a[1]) ? $a[1] : ""));
}
}
if ($minimal || $minimalp) {
parent::getWriteFile()->write($min);
} else {
parent::getWriteFile()->write($buf);
}
$min = '';
$buf = '';
$header = '';
}
//if(isset($intersection_of)) $buf .= $intersection_of.")].".PHP_EOL;
//if(isset($relationship)) $buf .= $relationship.")].".PHP_EOL;
if ($minimal || $minimalp) {
parent::getWriteFile()->Write($min);
} else {
parent::getWriteFile()->write($buf);
}
}