Skip to content

Commit

Permalink
Merge pull request #382 from Steinbeck-Lab/feat-search-api
Browse files Browse the repository at this point in the history 
Feat search api
  • Loading branch information
@CS76
CS76 authored Feb 7, 2025
2 parents 843f332 + 4783c82 commit 21996d9
Show file tree
Hide file tree
Showing 11 changed files with 279 additions and 2,395 deletions.
17 changes: 9 additions & 8 deletions app/Actions/Coconut/SearchMolecule.php
View file
Original file line number Diff line number Diff line change
Expand Up @@ -78,7 +78,7 @@ public function query($query, $size, $type, $sort, $tagType, $page)
}
}

return [$results, $this->collection, $this->organisms];
return [$results, $this->collection, $this->organisms, $this->citations];
} catch (QueryException $exception) {

return $this->handleException($exception);
Expand Down Expand Up @@ -202,7 +202,7 @@ private function buildTagsStatement($offset)
$query_organisms = array_map('strtolower', array_map('trim', explode(',', $this->query)));
$this->organisms = Organism::where(function ($query) use ($query_organisms) {
foreach ($query_organisms as $name) {
$query->orWhereRaw('LOWER(name) LIKE ?', ['%'.strtolower($name).'%']);
$query->orWhereRaw('name ILIKE ?', ['%'.$name.'%']);
}
})->get();
$organismIds = $this->organisms->pluck('id');
Expand All @@ -211,13 +211,14 @@ private function buildTagsStatement($offset)
$query->whereIn('organism_id', $organismIds);
})->where('active', true)->where('is_parent', false)->orderBy('annotation_level', 'DESC')->paginate($this->size);
} elseif ($this->tagType == 'citations') {
$this->citations = array_map('strtolower', array_map('trim', explode(',', $this->query)));
$citationIds = Citation::where(function ($query) {
foreach ($this->citations as $name) {
$query->orWhereRaw('LOWER(doi) LIKE ?', ['%'.strtolower($name).'%'])
->orWhereRaw('LOWER(title) LIKE ?', ['%'.strtolower($name).'%']);
$query_citations = array_map('strtolower', array_map('trim', explode(',', $this->query)));
$this->citations = Citation::where(function ($query) use ($query_citations) {
foreach ($query_citations as $name) {
$query->orWhereRaw('doi ILIKE ?', ['%'.$name.'%'])
->orWhereRaw('title ILIKE ?', ['%'.$name.'%']);
}
})->pluck('id');
})->get();
$citationIds = $this->citations->pluck('id');

return Molecule::whereHas('citations', function ($query) use ($citationIds) {
$query->whereIn('citation_id', $citationIds);
Expand Down
44 changes: 0 additions & 44 deletions app/Http/Controllers/API/CompoundController.php
View file

This file was deleted.

294 changes: 24 additions & 270 deletions app/Http/Controllers/API/SearchController.php
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2,291 +2,45 @@

namespace App\Http\Controllers\API;

use App\Actions\Coconut\SearchMolecule;
use App\Http\Controllers\Controller;
use App\Models\Molecule;
use Illuminate\Database\QueryException;
use Illuminate\Http\Request;
use Illuminate\Pagination\LengthAwarePaginator;
use Illuminate\Support\Facades\DB;
use Illuminate\Support\Facades\Cache;

class SearchController extends Controller
{
public function search(Request $request)
public function search(Request $request, SearchMolecule $search)
{
try {
set_time_limit(300);

$queryType = 'text';
$results = [];

$limit = $request->query('limit');
$sort = $request->query('sort');
$limit = $limit ? $limit : 24;
$page = $request->query('page');
$tagType = $request->get('tagType') ? $request->get('tagType') : null;

$offset =
(($page != null && $page != 'null' && $page != 0 ? $page : 1) -
1) *
$limit;

$query = $request->get('query');

$type = $request->query('type')
? $request->query('type')
: $request->get('type');

if ($type) {
$queryType = $type;
} else {
//inchi
$re =
'/^((InChI=)?[^J][0-9BCOHNSOPrIFla+\-\(\)\\\\\/,pqbtmsih]{6,})$/i';
preg_match_all($re, $query, $imatches, PREG_SET_ORDER, 0);

if (count($imatches) > 0 && substr($query, 0, 6) == 'InChI=') {
$queryType = 'inchi';
}

//inchikey
$re = '/^([0-9A-Z\-]+)$/i';
preg_match_all($re, $query, $ikmatches, PREG_SET_ORDER, 0);
if (
count($ikmatches) > 0 &&
substr($query, 14, 1) == '-' &&
strlen($query) == 27
) {
$queryType = 'inchikey';
}

// smiles
$re = '/^([^J][0-9BCOHNSOPrIFla@+\-\[\]\(\)\\\\\/%=#$]{6,})$/i';
preg_match_all($re, $query, $matches, PREG_SET_ORDER, 0);

if (count($matches) > 0 && substr($query, 14, 1) != '-') {
$queryType = 'smiles';
}
}

$filterMap = [
'mf' => 'molecular_formula',

'mw' => 'molecular_weight',
'hac' => 'heavy_atom_count',
'tac' => 'total_atom_count',
$query = $request->get('query');

'arc' => 'aromatic_ring_count',
'rbc' => 'rotatable_bond_count',
'mrc' => 'minimal_number_of_rings',
'fc' => 'formal_charge',
'cs' => 'contains_sugar',
'crs' => 'contains_ring_sugars',
'cls' => 'contains_linear_sugars',
$sort = $request->get('sort');
$type = $request->get('type') ? $request->get('type') : null;
$tagType = $request->get('tagType') ? $request->get('tagType') : null;
$page = $request->get('page');

'npl' => 'np_likeness_score',
'alogp' => 'alogp',
'topopsa' => 'topo_psa',
'fsp3' => 'fsp3',
'hba' => 'h_bond_acceptor_count',
'hbd' => 'h_bond_donor_count',
'ro5v' => 'rule_of_5_violations',
'lhba' => 'lipinski_h_bond_acceptor_count',
'lhbd' => 'lipinski_h_bond_donor_count',
'lro5v' => 'lipinski_rule_of_5_violations',
'ds' => 'found_in_databases',
$limit = $request->get('limit');
$limit = $limit ? $limit : 24;

'class' => 'chemical_class',
'subclass' => 'chemical_sub_class',
'superclass' => 'chemical_super_class',
'parent' => 'direct_parent_classification',
$results = [];

];
$offset = $request->query('offset');

$queryType = strtolower($queryType);

$statement = null;

if ($queryType == 'smiles' || $queryType == 'substructure') {
$statement =
"select id, COUNT(*) OVER () from mols where m@>'".
$query.
"' limit ".
$limit.
' offset '.
$offset;
} elseif ($queryType == 'inchi') {
$statement =
"select id, COUNT(*) OVER () from molecules where standard_inchi LIKE '%".
$query.
"%' limit ".
$limit.
' offset '.
$offset;
} elseif ($queryType == 'inchikey') {
$statement =
"select id, COUNT(*) OVER () from molecules where standard_inchi_key LIKE '%".
$query.
"%' limit ".
$limit.
' offset '.
$offset;
} elseif ($queryType == 'exact') {
$statement =
"select id, COUNT(*) OVER () from mols where m@='".
$query.
"' limit ".
$limit.
' offset '.
$offset;
} elseif ($queryType == 'similarity') {
$statement =
"select id, COUNT(*) OVER () from fps where mfp2%morganbv_fp('".
$query.
"') limit ".
$limit.
' offset '.
$offset;
} elseif ($queryType == 'tags') {
if ($tagType == 'dataSource') {

} else {
$results = Molecule::withAnyTags([$query], $tagType)->paginate($limit)->items();
$count = Molecule::withAnyTags([$query], $tagType)->count();
}
} elseif ($queryType == 'filters') {
$orConditions = explode('OR', $query);
$isORInitial = true;
$statement =
'select molecule_id as id, COUNT(*) OVER () from properties where ';
foreach ($orConditions as $orCondition) {
if ($isORInitial === false) {
$statement = $statement.' OR ';
}
$isORInitial = false;
$statement = $statement.'(';
$andConditions = explode(' ', trim($orCondition, ' '));
$isANDInitial = true;
foreach ($andConditions as $andCondition) {
if ($isANDInitial === false) {
$statement = $statement.' AND ';
}
$isANDInitial = false;
$_filter = explode(':', $andCondition);
if (str_contains($_filter[1], '..')) {
$range = array_values(explode('..', $_filter[1]));
$statement =
$statement.
'('.
$filterMap[$_filter[0]].
' between '.
$range[0].
' and '.
$range[1].
')';
} elseif (
$_filter[1] === 'true' ||
$_filter[1] === 'false'
) {
$statement =
$statement.
'('.
$filterMap[$_filter[0]].
' = '.
$_filter[1].
')';
} elseif (str_contains($_filter[1], '|')) {
$dbFilters = explode('|', $_filter[1]);
$dbs = explode('+', $dbFilters[0]);
$statement =
$statement.
'('.
$filterMap[$_filter[0]].
" @> '[\"".
implode('","', $dbs).
"\"]')";
} else {
if (str_contains($_filter[1], '+')) {
$_filter[1] = str_replace('+', ' ', $_filter[1]);
}
$statement =
$statement.
'('.$filterMap[$_filter[0]].'::TEXT ILIKE \'%'.$_filter[1].'%\')';
}
}
$statement = $statement.')';
}
$statement = $statement.' LIMIT '.$limit;
} else {
if ($query) {
$query = str_replace("'", "''", $query);
$statement =
"select id, COUNT(*) OVER () from molecules WHERE (\"name\"::TEXT ILIKE '%".
$query.
"%') OR (\"synonyms\"::TEXT ILIKE '%".
$query.
"%') OR (\"identifier\"::TEXT ILIKE '%".
$query.
"%') limit ".
$limit.
' offset '.
$offset;
} else {
$statement =
'select id, COUNT(*) OVER () from mols limit '.
$limit.
' offset '.
$offset.
'ORDER BY annotation_level DESC';
}
}
if ($statement) {
$expression = DB::raw($statement);
$qString = $expression->getValue(
DB::connection()->getQueryGrammar()
);
try {
$cacheKey = 'search.'.md5($query.$limit.$type.$sort.$tagType.$page);

$hits = DB::select($qString);
$results = Cache::remember($cacheKey, now()->addDay(), function () use ($search, $query, $limit, $type, $sort, $tagType, $page) {
return $search->query($query, $limit, $type, $sort, $tagType, $page);
});

$count = count($hits) > 0 ? $hits[0]->count : 0;
$collection = $results[1];
$organisms = $results[2];

$ids = implode(
',',
collect($hits)
->pluck('id')
->toArray()
);

if ($ids != '') {
$statement =
'SELECT * FROM molecules WHERE ID IN ('.
implode(
',',
collect($hits)
->pluck('id')
->toArray()
).
')';
if ($sort == 'recent') {
$statement = $statement.' ORDER BY created_at DESC';
}
$expression = DB::raw($statement);
$string = $expression->getValue(
DB::connection()->getQueryGrammar()
);
$results = DB::select($string);
} else {
$results = [];
$count = 0;
}
}
$pagination = new LengthAwarePaginator(
$results,
$count,
$limit,
$page
return response()->json(
[
'data' => $results[0],
],
200
);

return $pagination;
} catch (QueryException $exception) {
$message = $exception->getMessage();
if (str_contains(strtolower($message), strtolower('SQLSTATE[42P01]'))) {
Expand Down
Loading

0 comments on commit 21996d9

Please sign in to comment.