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();
}
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;
}
public function Run()
{
$file = "iproclass.tb.gz";
$ldir = $this->GetParameterValue('indir');
$odir = $this->GetParameterValue('outdir');
$rdir = $this->GetParameterValue('download_url');
$lfile = $ldir . $file;
if (!file_exists($lfile)) {
trigger_error($lfile . " not found. Will attempt to download.", E_USER_NOTICE);
parent::setParameterValue('download', true);
}
//download all files
$rfile = $rdir . $file;
if ($this->GetParameterValue('download') == true) {
echo "downloading {$file}... ";
utils::DownloadSingle($rfile, $lfile);
// $cmd = "gzip -c $lfile | split -d -l 1000000 --filter='gzip > $FILE.gz' - iproclass-"
}
$ofile = "iproclass.nq";
$gz = true;
parent::setReadFile($lfile, true);
echo "processing {$file}... ";
$this->process();
echo "done!" . PHP_EOL;
parent::getWriteFile()->close();
echo "generating dataset release file... ";
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("iProClass")->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z", filemtime($lfile)))->setFormat("text/tab-separated-value")->setFormat("application/gzip")->setPublisher("http://pir.georgetown.edu")->setHomepage("http://pir.georgetown.edu/iproclass")->setRights("use-share-modify")->setLicense("http://pir.georgetown.edu/pirwww/about/linkpir.shtml")->setDataset("http://identifiers.org/iproclass/");
$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($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/iproclass/iproclass.php")->setCreateDate($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() . $output_file->toRDF();
parent::setWriteFile($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');
if (parent::getParameterValue('omim_api_key') == '') {
$key_file = parent::getParameterValue('omim_api_key_file');
if (file_exists($key_file)) {
$key = trim(file_get_contents($key_file));
if ($key) {
parent::setParameterValue('omim_api_key', $key);
} else {
trigger_error("No API key found in the specified omim key file {$key_file}", E_USER_WARNING);
}
} else {
trigger_error("No OMIM key has been provided either by commmand line or in the expected omim key file {$key_file}", E_USER_WARNING);
}
}
// get the list of mim2gene entries
$entries = $this->GetListOfEntries($ldir);
// get the work specified
$list = trim(parent::getParameterValue('files'));
if ($list != 'all') {
// check if a hyphenated list was provided
if (($pos = strpos($list, "-")) !== FALSE) {
$start_range = substr($list, 0, $pos);
$end_range = substr($list, $pos + 1);
// get the whole list
$full_list = $this->GetListOfEntries($ldir);
// now intersect
foreach ($full_list as $e => $type) {
if ($e >= $start_range && $e <= $end_range) {
$myentries[$e] = $type;
}
}
$entries = $myentries;
} else {
// for comma separated list
$b = explode(",", parent::getParameterValue('files'));
foreach ($b as $e) {
$myentries[$e] = '';
}
$entries = array_intersect_key($entries, $myentries);
}
}
// set the write file
$gz = strstr(parent::getParameterValue('output_format'), ".gz") === FALSE ? false : true;
$outfile = 'omim.' . parent::getParameterValue('output_format');
parent::setWriteFile($odir . $outfile, $gz);
// declare the mapping method types
$this->get_method_type(null, true);
// iterate over the entries
$i = 0;
$total = count($entries);
foreach ($entries as $omim_id => $type) {
echo "processing " . ++$i . " of {$total} - omim# ";
$download_file = $ldir . $omim_id . ".json.gz";
$gzfile = "compress.zlib://{$download_file}";
// download if the file doesn't exist or we are told to
if (!file_exists($download_file) || parent::getParameterValue('download') == true) {
// download using the api
$url = parent::getParameterValue('omim_api_url') . '&apiKey=' . parent::getParameterValue('omim_api_key') . '&mimNumber=' . $omim_id;
$buf = file_get_contents($url);
if (strlen($buf) != 0) {
file_put_contents($download_file, $buf);
usleep(500000);
// limit of 4 requests per second
}
}
// load entry, parse and write to file
$entry = json_decode(file_get_contents($gzfile), true);
$omim_id = trim((string) $entry["omim"]["entryList"][0]["entry"]['mimNumber']);
echo $omim_id;
$this->ParseEntry($entry, $type);
parent::writeRDFBufferToWriteFile();
echo PHP_EOL;
}
parent::writeRDFBufferToWriteFile();
parent::getWriteFile()->close();
// generate the dataset description file
$source_file = (new DataResource($this))->setURI(parent::getParameterValue('omim_api_url'))->setTitle("OMIM " . parent::getDatasetVersion())->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z"))->setFormat("application/json")->setPublisher("http://omim.org")->setHomepage("http://omim.org")->setRights("use")->setRights("no-commercial")->setRights("registration-required")->setLicense("http://www.omim.org/help/agreement")->setDataset("http://identifiers.org/omim/");
$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 {$prefix} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/omim/omim.php")->setCreateDate($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() . $output_file->toRDF();
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
return TRUE;
}
function run()
{
$ldir = parent::getParameterValue('indir');
$odir = parent::getParameterValue('outdir');
// get the snps from pharmgkb
$snps = explode(",", parent::getParameterValue('files'));
if ($snps[0] == 'all') {
$snps = $this->getSNPs();
} else {
if ($snps[0] == 'clinical') {
$snps = $this->getSNPs(true);
} else {
if ($snps[0] == 'omim') {
$lfile = $ldir . 'snp_omimvar.txt';
if (!file_exists($lfile) || parent::getParameterValue('download') == true) {
$ret = utils::DownloadSingle('ftp://ftp.ncbi.nlm.nih.gov/snp/Entrez/snp_omimvar.txt', $lfile);
}
$snps = $this->processOMIMVar($lfile);
} else {
if ($snps[0] == 'pharmgkb') {
$lfile = $ldir . 'pharmgkb.snp.zip';
if (!file_exists($lfile) || parent::getParameterValue('download') == true) {
$ret = utils::DownloadSingle('http://www.pharmgkb.org/download.do?objId=rsid.zip&dlCls=common', $lfile);
}
$snps = $this->processPharmGKBSnps($lfile);
}
}
}
}
$outfile = $odir . "dbsnp." . parent::getParameterValue('output_format');
$gz = strstr(parent::getParameterValue('output_format'), ".gz") === FALSE ? false : true;
parent::setWriteFile($outfile, $gz);
$n = count($snps);
$z = 0;
foreach ($snps as $i => $snp) {
$file = $snp . '.xml.gz';
$infile = $ldir . $file;
$rfile = parent::getParameterValue('download_url') . $snp;
//$outfile = $odir.$snp.".".parent::getParameterValue('output_format');
// check if exists
$download = false;
if (!file_exists($infile)) {
//trigger_error($lfile." not found. Will attempt to download. ", E_USER_NOTICE);
parent::setParameterValue('download', true);
}
// download
if (parent::getParameterValue('download') == true) {
trigger_error("Downloading {$file}", E_USER_NOTICE);
$ret = utils::downloadSingle($rfile, "compress.zlib://" . $infile, true);
if ($ret === false) {
continue;
}
}
// process
echo "Processing {$snp} (" . ($i + 1) . "/{$n})" . PHP_EOL;
$this->parse($infile);
parent::writeRDFBufferToWriteFile();
if ($z++ % 10000 == 0) {
parent::clear();
}
}
parent::getWriteFile()->close();
// generate the dataset description file
$source_file = (new DataResource($this))->setURI($rfile)->setTitle("dbSNP " . parent::getDatasetVersion())->setRetrievedDate(date("Y-m-d\\TG:i:s\\Z"))->setFormat("application/xml")->setPublisher("http://www.ncbi.nlm.nih.gov")->setHomepage("http://www.ncbi.nlm.nih.gov/SNP/")->setRights("use-share-modify")->setLicense("http://www.ncbi.nlm.nih.gov/About/disclaimer.html")->setDataset("http://identifiers.org/dbsnp/");
$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 {$prefix} (generated at {$date})")->setSource($source_file->getURI())->setCreator("https://github.com/bio2rdf/bio2rdf-scripts/blob/master/dbsnp/dbsnp.php")->setCreateDate($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() . $output_file->toRDF();
parent::setWriteFile($odir . parent::getBio2RDFReleaseFile());
parent::getWriteFile()->write($dataset_description);
parent::getWriteFile()->close();
}