|
|
@@ -343,7 +343,7 @@ class OBOImporter extends TripalImporter {
|
|
|
$dfile = $_SERVER['DOCUMENT_ROOT'] . base_path() . $uobo_file;
|
|
|
if (!file_exists($dfile)) {
|
|
|
if (!file_exists($uobo_file)) {
|
|
|
- form_set_error('uobo_file', 'The specified path does not exist or cannot be read.');
|
|
|
+ form_set_error('uobo_file', t('The specified path, !path, does not exist or cannot be read.'), ['!path' => $dfile]);
|
|
|
}
|
|
|
}
|
|
|
if (!$uobo_url and !$uobo_file) {
|
|
|
@@ -367,7 +367,7 @@ class OBOImporter extends TripalImporter {
|
|
|
$dfile = $_SERVER['DOCUMENT_ROOT'] . base_path() . $obo_file;
|
|
|
if (!file_exists($dfile)) {
|
|
|
if (!file_exists($obo_file)) {
|
|
|
- form_set_error('obo_file', 'The specified path does not exist or cannot be read.');
|
|
|
+ form_set_error('obo_file', t('The specified path, !path, does not exist or cannot be read.'), ['!path' => $dfile]);
|
|
|
}
|
|
|
}
|
|
|
if (!$obo_url and !$obo_file) {
|
|
|
@@ -739,6 +739,156 @@ class OBOImporter extends TripalImporter {
|
|
|
$this->setItemsHandled($i);
|
|
|
return 1;
|
|
|
}
|
|
|
+
|
|
|
+ /**
|
|
|
+ * A helper function to get details about a foreign term.
|
|
|
+ *
|
|
|
+ * A foreign term is one that does not belong to the ontology.
|
|
|
+ *
|
|
|
+ * @param $t
|
|
|
+ * A term array that contains these keys at a minimum: id, name,
|
|
|
+ * definition, subset, namepace, is_obsolete.
|
|
|
+ * @param $default_db
|
|
|
+ * The default database.
|
|
|
+ */
|
|
|
+ private function _resolveTerm(&$t, $default_db) {
|
|
|
+
|
|
|
+ // We only check terms that have proper accessions (ie. short_name:accession)
|
|
|
+ if (strpos($t['id'], ':') === FALSE) {
|
|
|
+ return NULL;
|
|
|
+ }
|
|
|
+
|
|
|
+ // Get the short name and accession for the term.
|
|
|
+ $pair = explode(":", $t['id'], 2);
|
|
|
+ $short_name = $pair[0];
|
|
|
+ $accession_num = $pair[1];
|
|
|
+
|
|
|
+ // If the short name is the same as the deafult database then this
|
|
|
+ // term belongs in this ontology and should be defined in the file. We
|
|
|
+ // don't need to look it up.
|
|
|
+ if ($short_name == $default_db) {
|
|
|
+ return NULL;
|
|
|
+ }
|
|
|
+
|
|
|
+ // Before running the oboEbiLookup check for it in the local chado and
|
|
|
+ // the tripal_obo_temp table.
|
|
|
+ $sql = "
|
|
|
+ SELECT *
|
|
|
+ FROM {tripal_obo_temp} tot
|
|
|
+ WHERE tot.id = :term_id
|
|
|
+ ";
|
|
|
+ $result = chado_query($sql, array(':term_id' => $t['id'] . '_lookup'))->fetchObject();
|
|
|
+ $oterm = unserialize(base64_decode($result->stanza));
|
|
|
+ if ($oterm['label']) {
|
|
|
+ $t['name'] = $oterm['label'];
|
|
|
+ $t['definition'] = $oterm['def'];
|
|
|
+ $t['namepace'] = $oterm['ontology_name'];
|
|
|
+ $t['is_obsolete'] = $oterm['is_obsolete'];
|
|
|
+ $t['subset'] = $oterm['subset'];
|
|
|
+ $t['db_name'] = $oterm['ontology_prefix'];
|
|
|
+ return TRUE;
|
|
|
+ }
|
|
|
+
|
|
|
+ // Is the term in the local chado?
|
|
|
+ $sql = "SELECT * FROM {db} db WHERE db.name = :short_name ";
|
|
|
+ $db = chado_query($sql, array(':short_name' => $short_name))->fetchObject();
|
|
|
+ if (!empty($db)){
|
|
|
+ // Find the accession.
|
|
|
+ $sql = "
|
|
|
+ SELECT *
|
|
|
+ FROM {dbxref} dbx
|
|
|
+ WHERE dbx.db_id = :db_id
|
|
|
+ AND accession = :accession_num ";
|
|
|
+ $dbxref = chado_query($sql, array(':db_id' => $db->db_id, ':accession_num' => $accession_num))->fetchObject();
|
|
|
+ if (!empty($dbxref)) {
|
|
|
+ $sql = "
|
|
|
+ SELECT CVT.*, CV.name as namespace
|
|
|
+ FROM {cvterm} CVT
|
|
|
+ INNER JOIN {cv} CV on CV.cv_id = CVT.cv_id
|
|
|
+ WHERE dbxref_id = :dbxref_id
|
|
|
+ ";
|
|
|
+ $cvterm = chado_query($sql, [':dbxref_id' => $dbxref->dbxref_id])->fetchObject();
|
|
|
+ $t['name'] = $cvterm->name;
|
|
|
+ $t['definition'] = $cvterm->definition;
|
|
|
+ $t['namepace'] = $cvterm->namespace;
|
|
|
+ $t['is_obsolete'] = $cvterm->is_obsolete;
|
|
|
+ $t['db_name'] = $db->name;
|
|
|
+ return TRUE;
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ // Check for the ID of the term in EBI.
|
|
|
+ $oterm = NULL;
|
|
|
+ $results = $this->oboEbiLookup($t['id'], 'term');
|
|
|
+ if (isset($results['label'])) {
|
|
|
+ $oterm = $results;
|
|
|
+ }
|
|
|
+
|
|
|
+ // If we did not get a name for the term from a direct term
|
|
|
+ // lookup then let's try a query.,
|
|
|
+ if (!isset($results['label'])) {
|
|
|
+ $results = $this->oboEbiLookup($t['id'], 'query');
|
|
|
+ if (array_key_exists('docs', $results)) {
|
|
|
+ if (!empty($results['response']['docs'])) {
|
|
|
+ if (count($results['response']['docs']) > 1) {
|
|
|
+ foreach ($results['response']['docs'] as $doc) {
|
|
|
+ if ($doc['obo_id'] == $t['id']) {
|
|
|
+ $external = TRUE;
|
|
|
+ $oterm = $doc;
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+ else {
|
|
|
+ $external = true;
|
|
|
+ $oterm = $results['response']['docs'][0];
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ // If the accession could not be found in EBI.
|
|
|
+ if ($results['response']['numFound'] == 0 && !isset($results['label'])) {
|
|
|
+ // The first search doesn't work, so let's try a broader one.
|
|
|
+ $results = $this->oboEbiLookup($t['id'], 'query-non-local');
|
|
|
+ if (!empty($results)) {
|
|
|
+ if (array_key_exists('docs', $results)) {
|
|
|
+ if (!empty($results['docs'])) {
|
|
|
+ $accession = $t['id'];
|
|
|
+ $accession_underscore = str_replace(":", "_", $accession);
|
|
|
+ foreach ($results['response']['docs'] as $item) {
|
|
|
+ if ($item['label'] != $accession && $item['label'] != $accession_underscore) {
|
|
|
+ // Found the first place a label is other than the accession
|
|
|
+ // is used, so take that info and then end the loop.
|
|
|
+ $external = TRUE;
|
|
|
+ $oterm = $item;
|
|
|
+ break;
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ if ($oterm) {
|
|
|
+ $t['name'] = $oterm['label'];
|
|
|
+ $t['definition'] = $oterm['def'];
|
|
|
+ $t['namepace'] = $oterm['ontology_name'];
|
|
|
+ $t['is_obsolete'] = $oterm['is_obsolete'];
|
|
|
+ $t['subset'] = $oterm['subset'];
|
|
|
+ $t['db_name'] = $oterm['ontology_prefix'];
|
|
|
+
|
|
|
+ // Write the term to the tripal_obo_temp table for future reference
|
|
|
+ $values = array(
|
|
|
+ 'id' => $term_id . '_lookup',
|
|
|
+ 'stanza' => base64_encode(serialize($oterm)),
|
|
|
+ 'type' => 'lookup',
|
|
|
+ );
|
|
|
+ chado_insert_record('tripal_obo_temp', $values);
|
|
|
+ return TRUE;
|
|
|
+ }
|
|
|
+ // TODO: we couldn't find the term. What do we do!!!
|
|
|
+ return FALSE;
|
|
|
+ }
|
|
|
|
|
|
/**
|
|
|
* Uses the provided term array to add/update information to Chado about the
|
|
|
@@ -906,7 +1056,7 @@ class OBOImporter extends TripalImporter {
|
|
|
}
|
|
|
}
|
|
|
|
|
|
- $cvterm = tripal_insert_cvterm($t, array('update_existing' => TRUE));
|
|
|
+ $cvterm = chado_insert_cvterm($t, array('update_existing' => TRUE));
|
|
|
if (!$cvterm) {
|
|
|
throw new Exception("Cannot add the term " . $term['id'][0]);
|
|
|
}
|
|
|
@@ -1074,204 +1224,101 @@ class OBOImporter extends TripalImporter {
|
|
|
*/
|
|
|
private function addRelationship($cvterm, $defaultcv, $rel,
|
|
|
$objname, $object_is_relationship = 0, $default_db = 'OBO_REL') {
|
|
|
+
|
|
|
$reference_term = FALSE;
|
|
|
$in_obo = $this->getTerm($objname);
|
|
|
- // If an accession was passed we need to see if we can find the actual label.
|
|
|
- if (strpos($rel, ':') || strpos($objname, ':') && empty($in_obo['name'])) {
|
|
|
-
|
|
|
- if (strpos($rel, ':')) {
|
|
|
- $term_id = $rel;
|
|
|
- }
|
|
|
- elseif (strpos($objname, ':')) {
|
|
|
- $term_id = $objname;
|
|
|
- }
|
|
|
- $reference_term = TRUE;
|
|
|
- $pair = explode(":", $term_id, 2);
|
|
|
- $ontology_id = $pair[0];
|
|
|
- $accession_num = $pair[1];
|
|
|
- $rel_name = '';
|
|
|
-
|
|
|
- if (is_numeric($accession_num)) {
|
|
|
- // Before running the oboEbiLookup check for it in the local chado and
|
|
|
- // the tripal_obo_temp table.
|
|
|
- // $term_with_quotes = '"' . $term_id . '"';
|
|
|
- $sql = "
|
|
|
- SELECT *
|
|
|
- FROM {tripal_obo_temp} tot
|
|
|
- WHERE tot.id = :term_id
|
|
|
- ";
|
|
|
- $result = chado_query($sql, array(':term_id' => $term_id . '_lookup'))->fetchObject();
|
|
|
- $oterm = unserialize(base64_decode($result->stanza));
|
|
|
- if (empty($oterm['label'])){
|
|
|
-
|
|
|
- // Is this ontology is in the local chado?
|
|
|
- $sql = "SELECT * FROM {db} db WHERE db.name = :ontology_id ";
|
|
|
- $db = chado_query($sql, array(':ontology_id' => $ontology_id, ))->fetchObject();
|
|
|
-
|
|
|
- if (!empty($db)){
|
|
|
- // Find the accession.
|
|
|
- $sql = "
|
|
|
- SELECT *
|
|
|
- FROM {dbxref} dbx
|
|
|
- WHERE dbx.db_id = :db_id
|
|
|
- AND accession = :accession_num ";
|
|
|
- $v = chado_query($sql, array(':db_id' => $db->db_id, ':accession_num' => $accession_num))->fetchObject();
|
|
|
- if (!empty($dbxref)) {
|
|
|
- $sql = "SELECT * FROM {cvterm} WHERE dbxref_id = $dbxref->dbxref_id ";
|
|
|
- $oterm = chado_query($sql)->fetchObject();
|
|
|
- $rel_name = $oterm->name;
|
|
|
- }
|
|
|
- }
|
|
|
-
|
|
|
- if (empty($dbxref)){
|
|
|
- $results = $this->oboEbiLookup($term_id, 'term');
|
|
|
- if (isset($results['label'])) {
|
|
|
- $rel_name = $results['label'];
|
|
|
- $oterm = $results;
|
|
|
- }
|
|
|
- if (empty($rel_name)) {
|
|
|
- $results = $this->oboEbiLookup($term_id, 'query');
|
|
|
- if (array_key_exists('docs', $results['response'])) {
|
|
|
- if (!empty($results['response']['docs'])) {
|
|
|
- if (count($results['response']['docs']) > 1) {
|
|
|
- foreach ($results['response']['docs'] as $doc) {
|
|
|
- if ($doc['obo_id'] == $term_id) {
|
|
|
- $rel_name = $doc['label'];
|
|
|
- $oterm = $doc;
|
|
|
- }
|
|
|
- }
|
|
|
- } else {
|
|
|
- $rel_name = $results['response']['docs'][0]['label'];
|
|
|
- $oterm = $results['response']['docs'][0];
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
- if (empty($rel_name)) {
|
|
|
- // The first search doesn't work, so let's try a broader one.
|
|
|
- $results = $this->oboEbiLookup($term_id, 'query-non-local');
|
|
|
- if (!empty($results)) {
|
|
|
- if (array_key_exists('docs', $results['response'])) {
|
|
|
- if (!empty($results['response']['docs'])) {
|
|
|
- foreach ($results['response']['docs'] as $item) {
|
|
|
- if ($item['obo_id'] == $term_id) {
|
|
|
- //Found the first place a label is other than the accession is used, so take
|
|
|
- // that info and then end the loop.
|
|
|
- $rel_name = $item['label'];
|
|
|
- $oterm = $item;
|
|
|
- break;
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
- // Write the term to the tripal_obo_temp table for future reference
|
|
|
- $values = array(
|
|
|
- 'id' => $term_id . '_lookup',
|
|
|
- 'stanza' => base64_encode(serialize($oterm)),
|
|
|
- 'type' => 'lookup',
|
|
|
- );
|
|
|
- chado_insert_record('tripal_obo_temp', $values);
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
- // Make sure the relationship cvterm exists.
|
|
|
- $term = array(
|
|
|
- 'name' => $rel,
|
|
|
- 'id' => "$default_db:$rel",
|
|
|
- 'definition' => '',
|
|
|
- 'is_obsolete' => 0,
|
|
|
- 'cv_name' => $defaultcv,
|
|
|
- 'is_relationship' => TRUE,
|
|
|
- 'db_name' => $default_db
|
|
|
- );
|
|
|
- $relcvterm = tripal_insert_cvterm($term, array('update_existing' => FALSE));
|
|
|
-
|
|
|
- if (!$relcvterm) {
|
|
|
- // If the relationship term couldn't be found in the default_db provided
|
|
|
- // then do on more check to find it in the relationship ontology
|
|
|
- $term = array(
|
|
|
- 'name' => $rel,
|
|
|
- 'id' => "OBO_REL:$rel",
|
|
|
- 'definition' => '',
|
|
|
- 'is_obsolete' => 0,
|
|
|
- 'cv_name' => $defaultcv,
|
|
|
- 'is_relationship' => TRUE,
|
|
|
- 'db_name' => 'OBO_REL'
|
|
|
- );
|
|
|
- $relcvterm = tripal_insert_cvterm($term, array('update_existing' => FALSE));
|
|
|
+
|
|
|
+ // If the relationship term or the relationship object are defined using
|
|
|
+ // accession IDs then we need to break them apart and look them up to get
|
|
|
+ // the term name.
|
|
|
+ if (strpos($rel, ':') || strpos($objname, ':') && empty($in_obo['name'])) {
|
|
|
+
|
|
|
+ // Make sure the relationship cvterm exists.
|
|
|
+ $term = [];
|
|
|
+ $term['name'] = $rel;
|
|
|
+ $term['id'] = "$default_db:$rel";
|
|
|
+ $term['definition'] = '';
|
|
|
+ $term['is_obsolete'] = 0;
|
|
|
+ $term['cv_name'] = $defaultcv;
|
|
|
+ $term['is_relationship'] = TRUE;
|
|
|
+ $term['db_name'] = $default_db;
|
|
|
+
|
|
|
+ $this->_resolveTerm($term, $default_db);
|
|
|
+ print_r($term);
|
|
|
+
|
|
|
+ $relcvterm = chado_insert_cvterm($term, array('update_existing' => FALSE));
|
|
|
if (!$relcvterm) {
|
|
|
- throw new Exception("Cannot find the relationship term in the current ontology or in the relationship ontology: $rel\n");
|
|
|
- }
|
|
|
- }
|
|
|
-
|
|
|
- // Get the object term.
|
|
|
- if ($reference_term === TRUE && !empty($oterm)) {
|
|
|
- $objterm = array();
|
|
|
- $objterm['id'] = $oterm['obo_id'];
|
|
|
- $objterm['name'] = $oterm['label'];
|
|
|
- if (array_key_exists('def', $oterm)) {
|
|
|
- $objterm['definition'] = $oterm['def'];
|
|
|
- }
|
|
|
- if (array_key_exists('subset', $oterm)) {
|
|
|
- $objterm['subset'] = $oterm['subset'];
|
|
|
- }
|
|
|
- if (array_key_exists('namespace', $oterm)) {
|
|
|
- $objterm['namespace'] = $oterm['ontology_name'];
|
|
|
- }
|
|
|
- if (array_key_exists('is_obsolete', $oterm)) {
|
|
|
- $objterm['is_obsolete'] = $oterm['is_obsolete'];
|
|
|
- }
|
|
|
- }
|
|
|
- else {
|
|
|
- $oterm = $this->getTerm($objname);
|
|
|
- if (!$oterm) {
|
|
|
- throw new Exception("Could not find object term $objname\n");
|
|
|
- }
|
|
|
- $objterm = array();
|
|
|
- $objterm['id'] = $oterm['id'][0];
|
|
|
- $objterm['name'] = $oterm['name'][0];
|
|
|
- if (array_key_exists('def', $oterm)) {
|
|
|
- $objterm['definition'] = $oterm['def'][0];
|
|
|
+ // If the relationship term couldn't be found in the default_db provided
|
|
|
+ // then do one more check to find it in the relationship ontology
|
|
|
+ $term = array(
|
|
|
+ 'name' => $rel,
|
|
|
+ 'id' => "OBO_REL:$rel",
|
|
|
+ 'definition' => '',
|
|
|
+ 'is_obsolete' => 0,
|
|
|
+ 'cv_name' => $defaultcv,
|
|
|
+ 'is_relationship' => TRUE,
|
|
|
+ 'db_name' => 'OBO_REL'
|
|
|
+ );
|
|
|
+ $relcvterm = tripal_insert_cvterm($term, array('update_existing' => FALSE));
|
|
|
+ if (!$relcvterm) {
|
|
|
+ throw new Exception("Cannot find the relationship term in the current ontology or in the relationship ontology: $rel\n");
|
|
|
+ }
|
|
|
}
|
|
|
- if (array_key_exists('subset', $oterm)) {
|
|
|
- $objterm['subset'] = $oterm['subset'][0];
|
|
|
+
|
|
|
+ // Try to resolve the object term.
|
|
|
+ $objterm = ['id' => $objname];
|
|
|
+ $found = $this->_resolveTerm($objterm, $default_db);
|
|
|
+
|
|
|
+ // If the $found var is NULL it means the term does not need resolving as
|
|
|
+ // it's local to this ontology. In that case we can pull it from the
|
|
|
+ // obo lookup table.
|
|
|
+ if ($found === NULL) {
|
|
|
+ $oterm = $this->getTerm($objname);
|
|
|
+ if (!$oterm) {
|
|
|
+ throw new Exception("Could not find object term $objname\n");
|
|
|
+ }
|
|
|
+ $objterm = array();
|
|
|
+ $objterm['id'] = $oterm['id'][0];
|
|
|
+ $objterm['name'] = $oterm['name'][0];
|
|
|
+ if (array_key_exists('def', $oterm)) {
|
|
|
+ $objterm['definition'] = $oterm['def'][0];
|
|
|
+ }
|
|
|
+ if (array_key_exists('subset', $oterm)) {
|
|
|
+ $objterm['subset'] = $oterm['subset'][0];
|
|
|
+ }
|
|
|
+ if (array_key_exists('namespace', $oterm)) {
|
|
|
+ $objterm['namespace'] = $oterm['namespace'][0];
|
|
|
+ }
|
|
|
+ if (array_key_exists('is_obsolete', $oterm)) {
|
|
|
+ $objterm['is_obsolete'] = $oterm['is_obsolete'][0];
|
|
|
+ }
|
|
|
+ $t['db_name'] = $oterm['ontology_prefix'];
|
|
|
}
|
|
|
- if (array_key_exists('namespace', $oterm)) {
|
|
|
- $objterm['namespace'] = $oterm['namespace'][0];
|
|
|
+
|
|
|
+ $objterm['is_relationship'] = $object_is_relationship;
|
|
|
+
|
|
|
+ $objcvterm = chado_insert_cvterm($objterm, array('update_existing' => TRUE));
|
|
|
+ if (!$objcvterm) {
|
|
|
+ throw new Exception("Cannot add cvterm '" . $objterm['name'] . "' (" . $objterm['id'] . ").");
|
|
|
}
|
|
|
- if (array_key_exists('is_obsolete', $oterm)) {
|
|
|
- $objterm['is_obsolete'] = $oterm['is_obsolete'][0];
|
|
|
+
|
|
|
+ // check to see if the cvterm_relationship already exists, if not add it
|
|
|
+ $values = array(
|
|
|
+ 'type_id' => $relcvterm->cvterm_id,
|
|
|
+ 'subject_id' => $cvterm->cvterm_id,
|
|
|
+ 'object_id' => $objcvterm->cvterm_id
|
|
|
+ );
|
|
|
+ $result = chado_select_record('cvterm_relationship', array('*'), $values);
|
|
|
+ if (count($result) == 0) {
|
|
|
+ $options = array('return_record' => FALSE);
|
|
|
+ $success = chado_insert_record('cvterm_relationship', $values, $options);
|
|
|
+ if (!$success) {
|
|
|
+ throw new Exception("Cannot add term relationship: '$cvterm->name' $rel '$objcvterm->name'");
|
|
|
+ }
|
|
|
}
|
|
|
- }
|
|
|
-
|
|
|
- $objterm['cv_name' ] = $defaultcv;
|
|
|
- $objterm['is_relationship'] = $object_is_relationship;
|
|
|
- $objterm['db_name'] = $default_db;
|
|
|
|
|
|
- $objcvterm = chado_insert_cvterm($objterm, array('update_existing' => TRUE));
|
|
|
- if (!$objcvterm) {
|
|
|
- throw new Exception("Cannot add cvterm '" . $objterm['name'] . "' (" . $objterm['id'] . ").");
|
|
|
- }
|
|
|
-
|
|
|
- // check to see if the cvterm_relationship already exists, if not add it
|
|
|
- $values = array(
|
|
|
- 'type_id' => $relcvterm->cvterm_id,
|
|
|
- 'subject_id' => $cvterm->cvterm_id,
|
|
|
- 'object_id' => $objcvterm->cvterm_id
|
|
|
- );
|
|
|
- $result = chado_select_record('cvterm_relationship', array('*'), $values);
|
|
|
- if (count($result) == 0) {
|
|
|
- $options = array('return_record' => FALSE);
|
|
|
- $success = chado_insert_record('cvterm_relationship', $values, $options);
|
|
|
- if (!$success) {
|
|
|
- throw new Exception("Cannot add term relationship: '$cvterm->name' $rel '$objcvterm->name'");
|
|
|
- }
|
|
|
+ return TRUE;
|
|
|
}
|
|
|
-
|
|
|
- return TRUE;
|
|
|
}
|
|
|
|
|
|
/**
|
Stephen Ficklin