if ($sChromosome == 'M') {
// Instead of the UD, we just use the NC, it's small enough.
$sRefseqUD = $_SETT['human_builds'][$_CONF['refseq_build']]['ncbi_sequences'][$sChromosome];
} else {
// Get UD from mutalyzer.
try {
$sRefseqUD = lovd_getUDForGene($_CONF['refseq_build'], $sSymbol);
if ($sRefseqUD === '') {
// Function may return an empty string. This is not a SOAP error, but still an error. For instance a type of gene we don't support.
// To prevent further problems (getting transcripts, let's handle this nicely, shall we?
$_BAR->setMessage('Failed to retrieve gene reference sequence. This could be a temporary error, but it is likely that this gene is not supported by LOVD.', 'done');
$_BAR->setMessageVisibility('done', true);
die('</BODY>' . "\n" . '</HTML>' . "\n");
}
} catch (SoapFault $e) {
lovd_soapError($e);
}
}
// Get all transcripts and info.
// FIXME: When changing code here, check in transcripts?create if you need to make changes there, too.
$_BAR->setMessage('Collecting all available transcripts...');
$_BAR->setProgress($nProgress += 17);
$aTranscripts = $_DATA['Transcript']->getTranscriptPositions($sRefseqUD, $sSymbol, $sGeneName, $nProgress);
$_BAR->setProgress(100);
$_BAR->setMessage('Information collected, now building form...');
$_BAR->setMessageVisibility('done', true);
$_SESSION['work'][$sPath][$_POST['workID']]['step'] = '2';
$_SESSION['work'][$sPath][$_POST['workID']]['values'] = array('id' => $sSymbol, 'name' => $sGeneName, 'chromosome' => $sChromosome, 'chrom_band' => $sChromBand, 'id_hgnc' => $sHgncID, 'id_entrez' => $sEntrez, 'id_omim' => $nOmim, 'genomic_references' => $aRefseqGenomic, 'refseq_UD' => $sRefseqUD);
if (!empty($aTranscripts)) {
$_SESSION['work'][$sPath][$_POST['workID']]['values'] = array_merge($_SESSION['work'][$sPath][$_POST['workID']]['values'], array('transcripts' => $aTranscripts['id'], 'transcriptMutalyzer' => $aTranscripts['mutalyzer'], 'transcriptsProtein' => $aTranscripts['protein'], 'transcriptNames' => $aTranscripts['name'], 'transcriptPositions' => $aTranscripts['positions']));
}
/**
* This method returns transcripts and info from mutalyzer.
* Note that transcripts that are already in the LOVD database are skipped.
* @param string $sRefseqUD Genomic reference.
* @param string $sSymbol HGNC gene symbol.
* @param string $sGeneName HGNC gene name.
* @param float $nProgress Variable is passed by reference and used to keep up the progress of the progress bar.
* If the progress bar is initialized before this method is called, you can keep track of the progress with this variable.
* The progress bar will start at zero when this variable is not set.
* @return array $aTranscriptInfo Transcript information from mutalyzer.
**/
public function getTranscriptPositions($sRefseqUD, $sSymbol, $sGeneName, &$nProgress = 0.0)
{
global $_BAR, $_SETT, $_DB;
$_Mutalyzer = new LOVD_SoapClient();
$aTranscripts = array('id' => array(), 'name' => array(), 'mutalyzer' => array(), 'positions' => array(), 'protein' => array(), 'added' => array());
$sAliasSymbol = $sSymbol;
$aTranscripts['added'] = $_DB->query('SELECT id_ncbi FROM ' . TABLE_TRANSCRIPTS . ' WHERE geneid = ? ORDER BY id_ncbi', array($sSymbol))->fetchAllColumn();
if (isset($_SETT['mito_genes_aliases'][$sSymbol])) {
// For mitochondrial genes, an alias must be used to get the transcripts and info.
// List of aliases are hard-coded in inc-init.php.
$sAliasSymbol = $_SETT['mito_genes_aliases'][$sSymbol];
}
try {
// Can throw notice when TranscriptInfo is not present (when a gene recently has been renamed, for instance).
$aTranscripts['info'] = @$_Mutalyzer->getTranscriptsAndInfo(array('genomicReference' => $sRefseqUD, 'geneName' => $sAliasSymbol))->getTranscriptsAndInfoResult->TranscriptInfo;
} catch (SoapFault $e) {
lovd_soapError($e);
}
if (empty($aTranscripts['info'])) {
// No transcripts found.
$aTranscripts['info'] = array();
return $aTranscripts;
}
$nTranscripts = count($aTranscripts['info']);
foreach ($aTranscripts['info'] as $oTranscript) {
$nProgress += (100 - $nProgress) / $nTranscripts;
$_BAR->setMessage('Collecting ' . $oTranscript->id . ' info...');
if (isset($_SETT['mito_genes_aliases'][$sSymbol])) {
// For mitochondrial genes, we won't be able to get any proper transcript information. Fake one.
// FIXME: This code only works, if there is just one transcript. Can we assume there is only one?
// Perhaps it's better to use the same array construction as for normal genes, which is shorter, faster, and more flexible.
$sRefseqNM = $sRefseqUD . '(' . $sAliasSymbol . '_v001)';
if (in_array($sRefseqNM, $aTranscripts['added'])) {
// Transcript already exists; continue to the next transcript.
continue;
}
$aTranscripts['id'] = array($sRefseqNM);
$aTranscripts['protein'] = array($sRefseqNM => '');
// Until revision 679 the transcript version was not used in the index. The version number was removed with preg_replace.
// Can not figure out why version is not included. Therefore, for now we will do without preg_replace.
$aTranscripts['mutalyzer'] = array($sRefseqNM => '001');
$aTranscripts['name'] = array($sRefseqNM => 'transcript variant 1');
// FIXME: Perhaps indicate this transcript is a fake one, reconstructed from the CDS?
$aTranscripts['positions'] = array($sRefseqNM => array('chromTransStart' => isset($oTranscript->gTransStart) ? $oTranscript->gTransStart : 0, 'chromTransEnd' => isset($oTranscript->gTransEnd) ? $oTranscript->gTransEnd : 0, 'cTransStart' => isset($oTranscript->cTransStart) ? $oTranscript->cTransStart : 0, 'cTransEnd' => isset($oTranscript->sortableTransEnd) ? $oTranscript->sortableTransEnd : 0, 'cCDSStop' => isset($oTranscript->cCDSStop) ? $oTranscript->cCDSStop : 0));
} else {
if (in_array($oTranscript->id, $aTranscripts['added'])) {
// Transcript already exists; continue to the next transcript.
continue;
}
$aTranscripts['id'][] = $oTranscript->id;
// Until revision 679 the transcript version was not used in the index. The version number was removed with preg_replace.
// Can not figure out why version is not included. Therefore, for now we will do without preg_replace.
$aTranscripts['name'][$oTranscript->id] = str_replace($sGeneName . ', ', '', $oTranscript->product);
$aTranscripts['mutalyzer'][$oTranscript->id] = str_replace($sSymbol . '_v', '', $oTranscript->name);
$aTranscripts['positions'][$oTranscript->id] = array('chromTransStart' => isset($oTranscript->chromTransStart) ? $oTranscript->chromTransStart : 0, 'chromTransEnd' => isset($oTranscript->chromTransEnd) ? $oTranscript->chromTransEnd : 0, 'cTransStart' => $oTranscript->cTransStart, 'cTransEnd' => $oTranscript->sortableTransEnd, 'cCDSStop' => $oTranscript->cCDSStop);
$aTranscripts['protein'][$oTranscript->id] = !isset($oTranscript->proteinTranscript) ? '' : $oTranscript->proteinTranscript->id;
}
$_BAR->setProgress($nProgress);
}
return $aTranscripts;
}
