Skip to content

Commit

Permalink
Merge pull request #1328 from tripal/1327-tv3-lookup-organism-from-name
Browse files Browse the repository at this point in the history 
Implement chado_get_organism_id_from_scientific_name
  • Loading branch information
@spficklin
spficklin committed Feb 13, 2023
2 parents 9c308ce + 7425c05 commit 8fd59e9
Showing 1 changed file with 150 additions and 2 deletions.
152 changes: 150 additions & 2 deletions tripal_chado/api/modules/tripal_chado.organism.api.inc
View file
Original file line number Diff line number Diff line change
Expand Up @@ -21,8 +21,9 @@
* An array with the key stating what the identifier is. Supported keys (only
* one of the following unique keys is required):
* - organism_id: the chado organism.organism_id primary key.
* - genus & species: the chado organism.genus field & organism.species
* field. There are also some specially handled keys. They are:
* - genus & species: the chado organism.genus field & organism.species field.
* - scientific_name: Full taxonomic name, can include infraspecific nomenclature.
* There are also some specially handled keys. They are:
* - property: An array/object describing the property to select records
* for.
* It should at least have either a type_name (if unique across cvs) or
Expand Down Expand Up @@ -77,6 +78,28 @@ function chado_get_organism($identifiers, $options = []) {
);
}

// If the scientific_name identifier is used, we look up organism_id from that.
if (isset($identifiers['scientific_name'])) {
$scientific_name = $identifiers['scientific_name'];
unset($identifiers['scientific_name']);
$organism_ids = chado_get_organism_id_from_scientific_name($scientific_name, $options);
if (count($organism_ids) == 1) {
$identifiers['organism_id'] = $organism_ids[0];
}
else {
tripal_report_error(
'tripal_organism_api',
TRIPAL_ERROR,
"chado_get_organism: The specified scientific name did not uniquely identify an organism.
You passed in %scientific_name.",
[
'%scientific_name' => $scientific_name,
]
);
return NULL;
}
}

// If one of the identifiers is property then use chado_get_record_with_property().
if (isset($identifiers['property'])) {
$property = $identifiers['property'];
Expand Down Expand Up @@ -181,6 +204,131 @@ function chado_get_organism_scientific_name($organism) {
return $name;
}

/**
* Returns organism_id values of organisms matching the specified full
* scientific name, abbreviation, or common name of an organism.
*
* @param $name
* The organism name to be queried. Infraspecific type can be abbreviated.
*
* @param $options
* An array of options. The following keys are available:
* - check_abbreviation: If TRUE and the $name did not match the
* scientific name, then check the abbreviation.
* - check_common_name: If TRUE and the $name did not match the
* scientific name, then check the common name.
* - case_sensitive: If TRUE then all searches should be case
* sensitive. Default is FALSE.
* If no options are specified, search is for a match of $name to
* the scientific_name only, case insensitive.
*
* @return
* Array of matching organism_id values.
*
* @ingroup tripal_organism_api
*/
function chado_get_organism_id_from_scientific_name($name, $options = []) {
$organism_ids = [];

// Handle missing $name by returning empty array.
if (!$name) {
return $organism_ids;
}

// By default, search is case insensitive because this function may
// be used to handle input from users or from data files in loaders.
$sql_for_lower = '';
if (!in_array('case_sensitive', $options, true)) {
$name = strtolower($name);
$sql_for_lower = 'LOWER';
}

// Check scientific name first, and if a match is found, nothing
// else specified by $options will be checked.
// Scientific name is the combination of genus, species,
// and optionally infraspecific nomenclature added with Chado 1.3
// For Chado 1.2 and earlier, infraspecific nomenclature had to
// be stored in the species column, so limit the split accordingly.
// There is a unique constraint, so expect zero or one match here.
$infra_present = false;
$limit = 2;
if (chado_column_exists('organism', 'infraspecific_name')) {
$infra_present = true;
$limit = 4;
}
$parts = preg_split('/\s+/', $name, $limit);
// $name could be a single word, so make sure this is defined.
if (!array_key_exists(1, $parts)) {
$parts[1] = '';
}
$sql = 'SELECT organism_id FROM {organism} WHERE '.$sql_for_lower.'(genus) = :genus'
. ' AND '.$sql_for_lower.'(species) = :species';
$args = [ ':genus' => $parts[0], ':species' => $parts[1] ];
if ($infra_present) {
// When there is no infraspecific name, we can either use the "no_rank"
// taxonomic term in the type_id column, or else use NULL.
$sql .= ' AND ( type_id = (SELECT cvterm_id FROM {cvterm}'
. ' WHERE '.$sql_for_lower.'(name) = :infraspecific_type'
. ' AND cv_id = (SELECT cv_id FROM {cv} WHERE name = :taxonomic_rank))';
if (!array_key_exists(2, $parts)) {
$parts[2] = 'no_rank';
$sql .= " OR type_id IS NULL";
}
else {
$parts[2] = chado_unabbreviate_infraspecific_rank($parts[2]);
}
$sql .= ")";
$args[':infraspecific_type'] = $parts[2];
$args[':taxonomic_rank'] = 'taxonomic_rank';

// Infraspecific name, if present.
if (array_key_exists(3, $parts)) {
$sql .= ' AND '.$sql_for_lower.'(infraspecific_name) = :infraspecific_name';
$args[':infraspecific_name'] = $parts[3];
}
else {
// Infraspecific name not present, so this column
// must be either an empty string or NULL.
$sql .= " AND ( infraspecific_name = '' ) IS NOT FALSE";
}
}
$results = chado_query($sql, $args);
while ($organism = $results->fetchField()) {
if (!in_array($organism, $organism_ids)) {
$organism_ids[] = $organism;
}
}

// Check other search modes only when no match was found for scientific name.
if (empty($organism_ids)) {
// Try to find $name in the abbreviation column. This does not
// have a unique constraint, so there may be more than one match.
if (in_array('check_abbreviation', $options, true)) {
$sql = 'SELECT organism_id FROM {organism} WHERE '.$sql_for_lower.'(abbreviation) = :name';
$args = [':name' => $name];
$results = chado_query($sql, $args);
while ($organism = $results->fetchField()) {
$organism_ids[] = $organism;
}
}

// Try to find $name in the common_name column. This does not
// have a unique constraint, so there may be more than one match.
if (in_array('check_common_name', $options, true)) {
$sql = 'SELECT organism_id FROM {organism} WHERE '.$sql_for_lower.'(common_name) = :name';
$args = [':name' => $name];
$results = chado_query($sql, $args);
while ($organism = $results->fetchField()) {
if (!in_array($organism, $organism_ids)) {
$organism_ids[] = $organism;
}
}
}
}

return $organism_ids;
}

/**
* Returns a list of organisms to use in select lists.
*
Expand Down

0 comments on commit 8fd59e9

Please sign in to comment.