function lovd_getGeneInfoFromHGNC($sHgncId, $bRecursion = false)
{
// Uses the HGNC API to search and retrieve gene information from the HGNC
// website. The first argument can be an HGNC accession number or an HGNC
// approved gene symbol. The results will be returned as an associative
// array. If $bRecursion == true, this function automatically handles
// deprecated HGNC entries and returns the information with the correct gene
// symbol. On error, this function calls lovd_errorAdd if inc-lib-form.php
// was included. It always returns false on failure.
// 2014-01-13; 3.0-09; Tired of problems when their URL change and they don't announce it properly; rewrite to use their new REST service.
$sURL = 'http://rest.genenames.org/search/';
if (ctype_digit($sHgncId)) {
$sURL .= 'hgnc_id/';
} else {
$sURL .= 'symbol/';
}
$sURL .= $sHgncId;
$nHGNCID = 0;
$aOutput = lovd_php_file($sURL, false, '', 'Accept: application/json');
if ($aOutput && ($aOutput = json_decode(implode('', $aOutput), true))) {
if (!empty($aOutput['response']['numFound'])) {
// 2014-08-06; 3.0-11; HGNC *again* changed their output, and once again we need to adapt quickly.
$nHGNCID = preg_replace('/[^0-9]+/', '', $aOutput['response']['docs'][0]['hgnc_id']);
} else {
// Not found, previous symbol of...?
$sURL = str_replace('/symbol/', '/prev_symbol/', $sURL);
$aOutput = lovd_php_file($sURL, false, '', 'Accept: application/json');
if ($aOutput && ($aOutput = json_decode(implode('', $aOutput), true))) {
if (!empty($aOutput['response']['numFound'])) {
if ($aOutput['response']['numFound'] == 1 && $bRecursion) {
// 2014-08-06; 3.0-11; HGNC *again* changed their output, and once again we need to adapt quickly.
$nHGNCID = preg_replace('/[^0-9]+/', '', $aOutput['response']['docs'][0]['hgnc_id']);
return lovd_getGeneInfoFromHGNC($nHGNCID, $bRecursion);
} elseif (function_exists('lovd_errorAdd')) {
$sSymbols = '';
for ($i = 0; $i < $aOutput['response']['numFound']; $i++) {
$sSymbols .= (!$i ? '' : ($i == $aOutput['response']['numFound'] - 1 ? ' or ' : ', ')) . $aOutput['response']['docs'][$i]['symbol'];
}
lovd_errorAdd('hgnc_id', 'Entry ' . htmlspecialchars($sHgncId) . ' is deprecated according to the HGNC, please use ' . $sSymbols . '.');
}
return false;
} else {
// Not found, maybe it's an alias?
$sURL = str_replace('/prev_symbol/', '/alias_symbol/', $sURL);
$aOutput = lovd_php_file($sURL, false, '', 'Accept: application/json');
if ($aOutput && ($aOutput = json_decode(implode('', $aOutput), true))) {
if (!empty($aOutput['response']['numFound'])) {
if ($aOutput['response']['numFound'] == 1 && $bRecursion) {
// 2014-08-06; 3.0-11; HGNC *again* changed their output, and once again we need to adapt quickly.
$nHGNCID = preg_replace('/[^0-9]+/', '', $aOutput['response']['docs'][0]['hgnc_id']);
return lovd_getGeneInfoFromHGNC($nHGNCID, $bRecursion);
} elseif (function_exists('lovd_errorAdd')) {
$sSymbols = '';
for ($i = 0; $i < $aOutput['response']['numFound']; $i++) {
$sSymbols .= (!$i ? '' : ($i == $aOutput['response']['numFound'] - 1 ? ' or ' : ', ')) . $aOutput['response']['docs'][$i]['symbol'];
}
lovd_errorAdd('hgnc_id', 'Entry ' . htmlspecialchars($sHgncId) . ' was not found, perhaps you are referring to ' . $sSymbols . '.');
}
return false;
}
}
}
}
}
} else {
if (function_exists('lovd_errorAdd')) {
lovd_errorAdd('', 'Couldn\'t search for gene, probably because the HGNC is having website or database problems. Please try again later.');
}
return false;
}
if (!$nHGNCID) {
if (function_exists('lovd_errorAdd')) {
lovd_errorAdd('hgnc_id', 'Entry ' . htmlspecialchars($sHgncId) . ' was not found in the HGNC database.');
}
return false;
}
// Now that we have an ID, fetch the data. Use HGNC's fetch API.
$aOutput = lovd_php_file('http://rest.genenames.org/fetch/hgnc_id/' . $nHGNCID, false, '', 'Accept: application/json');
if ($aOutput && ($aOutput = json_decode(implode('', $aOutput), true))) {
if (!empty($aOutput['response']['numFound'])) {
$aGene = $aOutput['response']['docs'][0];
} else {
if (function_exists('lovd_errorAdd')) {
lovd_errorAdd('hgnc_id', 'Couldn\'t fetch gene information, even though search results were returned. This might be a problem at the side of the HGNC, or a bug in LOVD. Please try again later.');
}
return false;
}
} else {
if (function_exists('lovd_errorAdd')) {
lovd_errorAdd('', 'Couldn\'t get gene information, probably because the HGNC is having website or database problems. Please try again later.');
}
return false;
}
// Check returned entry.
// We ignore genes from the following locus groups:
$aBadLocusGroups = array('phenotype', 'withdrawn');
// We ignore genes from the following locus types (most of these are in group "other"):
$aBadLocusTypes = array('endogenous retrovirus', 'fragile site', 'immunoglobulin gene', 'region', 'transposable element', 'unknown', 'virus integration site', 'immunoglobulin pseudogene');
if ($aGene['status'] != 'Approved') {
if (function_exists('lovd_errorAdd')) {
lovd_errorAdd('hgnc_id', 'Entry ' . htmlspecialchars($sHgncId) . ' is not an approved gene symbol according to the HGNC database.');
}
return false;
} elseif ($aGene['location'] == 'reserved') {
if (function_exists('lovd_errorAdd')) {
lovd_errorAdd('hgnc_id', 'Entry ' . htmlspecialchars($sHgncId) . ' does not yet have a public association with a chromosomal location.');
}
return false;
} elseif (in_array($aGene['locus_group'], $aBadLocusGroups)) {
if (function_exists('lovd_errorAdd')) {
lovd_errorAdd('hgnc_id', 'LOVD cannot process this type of gene entry ' . htmlspecialchars($sHgncId) . ' (Locus Group: ' . $aGene['locus_group'] . ').');
}
return false;
} elseif (in_array($aGene['locus_type'], $aBadLocusTypes)) {
if (function_exists('lovd_errorAdd')) {
lovd_errorAdd('hgnc_id', 'LOVD cannot process this type of gene entry ' . htmlspecialchars($sHgncId) . ' (Locus Type: ' . $aGene['locus_type'] . ').');
}
return false;
}
foreach (array('omim_id') as $sCol) {
// Columns presented as arrays (new?), but should contain just one value.
// 2014-12-23; 3.0-13; Can also not be defined.
// 2016-09-14; 3.0-17; HGNC can actually return multiple OMIM IDs,
// take just the first one.
if (!isset($aGene[$sCol])) {
$aGene[$sCol] = '';
} elseif (is_array($aGene[$sCol])) {
$aGene[$sCol] = $aGene[$sCol][0];
}
}
// 2014-08-06; 3.0-11; HGNC ID suddenly got a prefix, removing the prefix.
$aGene['hgnc_id'] = preg_replace('/[^0-9]+/', '', $aGene['hgnc_id']);
// 2014-12-23; 3.0-13; Split "location" to "chromosome" and "chrom_band", which makes it easier to create this gene in the database.
if ($aGene['location'] == 'mitochondria') {
$aGene['chromosome'] = 'M';
$aGene['chrom_band'] = '';
} else {
preg_match('/^(\\d{1,2}|[XY])(.*)$/', $aGene['location'], $aMatches);
$aGene['chromosome'] = $aMatches[1];
$aGene['chrom_band'] = $aMatches[2];
}
return $aGene;
}
if (POST) {
lovd_errorClean();
if (empty($_POST['hgnc_id'])) {
lovd_errorAdd('hgnc_id', 'No HGNC ID or Gene symbol was specified');
} else {
// Gene Symbol must be unique.
// Enforced in the table, but we want to handle this gracefully.
// When numeric, we search the id_hgnc field. When not, we search the id (gene symbol) field.
$sSQL = 'SELECT id, id_hgnc FROM ' . TABLE_GENES . ' WHERE id' . (!ctype_digit($_POST['hgnc_id']) ? '' : '_hgnc') . ' = ?';
$aSQL = array($_POST['hgnc_id']);
$result = $_DB->query($sSQL, $aSQL)->fetchObject();
if ($result !== false) {
lovd_errorAdd('hgnc_id', sprintf('This gene entry (%s, HGNC-ID=%d) is already present in this LOVD installation.', $result->id, $result->id_hgnc));
} else {
// This call already makes the needed lovd_errorAdd() calls.
$aGeneInfo = lovd_getGeneInfoFromHGNC($_POST['hgnc_id']);
if (!empty($aGeneInfo)) {
$sHgncID = $aGeneInfo['hgnc_id'];
$sSymbol = $aGeneInfo['symbol'];
$sGeneName = $aGeneInfo['name'];
$sChromLocation = $aGeneInfo['location'];
$sEntrez = $aGeneInfo['entrez_id'];
$nOmim = $aGeneInfo['omim_id'];
}
}
}
if (!lovd_error()) {
$_T->printHeader();
$_T->printTitle();
require ROOT_PATH . 'class/progress_bar.php';
$sFormNextPage = '<FORM action="' . $sPath . '" id="createGene" method="post">' . "\n" . ' <INPUT type="hidden" name="workID" value="' . $_POST['workID'] . '">' . "\n" . ' <INPUT type="submit" value="Continue »">' . "\n" . ' </FORM>';
