tripal_cv.api.inc 27 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878
  1. <?php
  2. /**
  3. * @file
  4. * Controlled Vocabulary API
  5. *
  6. * @defgroup tripal_cv_api CV Module API
  7. * @ingroup tripal_api
  8. * This module provides a set of functions to simplify working with
  9. * controlled vocabularies. Most of the API functions deal with retrieving
  10. * terms or their parent vocabularies.
  11. *
  12. * However, the API also supports
  13. * generation of trees for browsing a vocabulary as well as generation of
  14. * pie graphs for display of hierarchical counts of terms. Version 0.3b of
  15. * Tripal provides a feature browser and a feature summary chart uses
  16. * the API functions provided here. But in general charts and trees can be
  17. * created for any controlled vocabulary.
  18. *
  19. */
  20. /**
  21. * Purpose: To retrieve a chado controlled vocabulary object
  22. *
  23. * @param $select_values
  24. * An array meant to uniquely select a given controlled vocabulary
  25. *
  26. * @return
  27. * Chado controlled vocabulary object
  28. *
  29. * The controlled vocabulary is selected using tripal_core_chado select and as such the
  30. * $select_values array parameter meant to uniquely identify the controlled vocab to be
  31. * returned follows the same form as when using tripal_core_chado_select directly.
  32. *
  33. * Example Usage:
  34. * @code
  35. $select_values = array(
  36. 'name' => 'feature_property'
  37. );
  38. $cv_object = tripal_cv_get_cv($select_values);
  39. * @endcode
  40. * The above code selects the feature_property cv and returns the following object:
  41. * @code
  42. $cv_object = stdClass Object (
  43. [cv_id] => 13
  44. [name] => feature_property
  45. [definition] =>
  46. );
  47. * @endcode
  48. *
  49. * @ingroup tripal_cv_api
  50. */
  51. function tripal_cv_get_cv($select_values) {
  52. $columns = array(
  53. 'cv_id',
  54. 'name',
  55. 'definition',
  56. );
  57. $results = tripal_core_chado_select('cv', $columns, $select_values);
  58. if (sizeof($results) == 1) {
  59. return $results[0];
  60. }
  61. elseif (empty($results)) {
  62. watchdog('tripal_cv',
  63. 'tripal_cv_get_cv: No cv matches criteria values:%values',
  64. array('%values' => print_r($select_values, TRUE)),
  65. WATCHDOG_WARNING
  66. );
  67. return FALSE;
  68. }
  69. else {
  70. watchdog('tripal_cv',
  71. 'tripal_cv_get_cv: 2+ cvs match criteria values:%values',
  72. array('%values' => print_r($select_values, TRUE)),
  73. WATCHDOG_WARNING
  74. );
  75. }
  76. }
  77. // Purpose: To retrieve a chado cv object
  78. // @param $where_options
  79. // @code
  80. // array(
  81. // <column_name> => array(
  82. // 'type' => <type of column: INT/STRING>,
  83. // 'value' => <the vlaue you want to filter on>,
  84. // 'exact' => <if TRUE use =; if FALSE use ~>,
  85. // )
  86. // )
  87. // @endcode
  88. //
  89. // @return
  90. // Chado cv object with all fields from the chado cv table
  91. //
  92. // @ingroup tripal_cv_api
  93. //
  94. //function tripal_cv_get_cv ($where_options)
  95. /**
  96. * Retrieve a cv given the cv name
  97. *
  98. * @param $name
  99. * The name of the cv to be returned
  100. * @return
  101. * The cv object for the specified CV name
  102. *
  103. * @ingroup tripal_cv_api
  104. */
  105. function tripal_cv_get_cv_by_name($name) {
  106. $r = tripal_core_chado_select('cv', array('*'), array('name' => $name));
  107. return $r[0];
  108. }
  109. /**
  110. * Retrieve the cv object for the specified CV id
  111. *
  112. * NOTE: This function is deprecated.
  113. * @see tripal_core_chado_generate_vars()
  114. *
  115. * @param $cv_id
  116. * The unique identifier for the cv to retrieve
  117. *
  118. * @return
  119. * An object describing the cv
  120. *
  121. * @ingroup tripal_cv_api
  122. */
  123. function tripal_cv_get_cv_by_id($cv_id) {
  124. $r = tripal_core_chado_select('cv', array('*'), array('cv_id' => $cv_id));
  125. return $r;
  126. }
  127. /**
  128. * Retrieve the cv id for the specified CV by name
  129. *
  130. * NOTE: This function is deprecated.
  131. * @see tripal_core_chado_generate_vars()
  132. *
  133. * @param $cv_name
  134. * The unique name for the cv to retrieve
  135. *
  136. * @return
  137. * The numeric cv ID
  138. *
  139. * @ingroup tripal_cv_api
  140. */
  141. function tripal_cv_get_cv_id($cv_name) {
  142. $sql = "SELECT cv_id FROM {cv} WHERE name = :name";
  143. $cv = chado_query($sql, array(':name' => $cv_name))->fetchObject();
  144. return $cv->cv_id;
  145. }
  146. /**
  147. * Create an options array to be used in a form element which provides a
  148. * list of all chado cvs
  149. *
  150. * NOTE: This function is deprecated as of Tripal v1.0
  151. *
  152. * @return
  153. * An array(cv_id => name) for each cv in the chado cv table
  154. *
  155. * @ingroup tripal_cv_api
  156. */
  157. function tripal_cv_get_cv_options() {
  158. $results = tripal_core_chado_select('cv', array('cv_id', 'name'), array());
  159. $options = array();
  160. foreach ($results as $r) {
  161. $options[$r->cv_id] = $r->name;
  162. }
  163. return $options;
  164. }
  165. /**
  166. * Retrieve a chado cvterm object with a given name
  167. *
  168. * @param $cvterm_id
  169. * the cvterm.cvterm_id
  170. *
  171. * @return
  172. * cvterm array or FALSE on error
  173. *
  174. * @ingroup tripal_cv_api
  175. */
  176. function tripal_cv_get_cvterm_by_id($cvterm_id) {
  177. if (!is_numeric($cvterm_id)) {
  178. return FALSE;
  179. }
  180. $values = array('cvterm_id' => $cvterm_id);
  181. $options = array('statement_name' => 'sel_cvterm_id');
  182. $r = tripal_core_chado_select('cvterm', array('*'), $values, $options);
  183. if (!$r) {
  184. return FALSE;
  185. }
  186. return $r[0];
  187. }
  188. /**
  189. * Retrieve a chado cvterm object with a given name
  190. *
  191. * @param $name
  192. * the cvterm.name
  193. * @param $cv_id
  194. * the cv_id of the term you are looking for
  195. * @param $cv_name
  196. * the name of the CV
  197. *
  198. * @return
  199. * cvterm array or FALSE on error
  200. *
  201. * @ingroup tripal_cv_api
  202. */
  203. function tripal_cv_get_cvterm_by_name($name, $cv_id = NULL, $cv_name = 'tripal') {
  204. if ($cv_id) {
  205. $values = array(
  206. 'name' => $name,
  207. 'cv_id' => $cv_id,
  208. );
  209. $options = array(
  210. 'statement_name' => 'sel_cvterm_nacv',
  211. 'case_insensitive_columns' => array('name')
  212. );
  213. $r = tripal_core_chado_select('cvterm', array('*'), $values, $options);
  214. }
  215. elseif ($cv_name) {
  216. $values = array(
  217. 'name' => $name,
  218. 'cv_id' => array(
  219. 'name' => $cv_name,
  220. ),
  221. );
  222. $options = array(
  223. 'statement_name' => 'sel_cvterm_nacv',
  224. 'case_insensitive_columns' => array('name')
  225. );
  226. $r = tripal_core_chado_select('cvterm', array('*'), $values, $options);
  227. }
  228. else {
  229. $values = array(
  230. 'name' => $name,
  231. );
  232. $options = array(
  233. 'statement_name' => 'sel_cvterm_na',
  234. 'case_insensitive_columns' => array('name')
  235. );
  236. $r = tripal_core_chado_select('cvterm', array('*'), $values, $options);
  237. }
  238. if (!$r) {
  239. return FALSE;
  240. }
  241. if (count($r) > 1) {
  242. watchdog('tripal_cv', "Cannot find a unique term for the term '%name' in the vocabulary '%cv'. Multiple entries exist for this name",
  243. array('%name' => $name, '%cv' => $cv_name ? $cv_name : $cv_id), WATCHDOG_ERROR);
  244. return FALSE;
  245. }
  246. return $r[0];
  247. }
  248. /**
  249. * Retrieve a chado cvterm object with a given name
  250. *
  251. * @param $synonym
  252. * the synonym of the term
  253. * @param $cv_id
  254. * the cv_id of the term you are looking for
  255. * @param $cv_name
  256. * the name of the CV
  257. *
  258. * @return
  259. * cvterm object
  260. *
  261. * @ingroup tripal_cv_api
  262. */
  263. function tripal_cv_get_cvterm_by_synonym($synonym, $cv_id = NULL, $cv_name = 'tripal') {
  264. // first find the CVTerm synonym
  265. $values = array(
  266. 'synonym' => $synonym,
  267. );
  268. $statement = "sel_cvtermsynonym_sy";
  269. if ($cv_id) {
  270. $values['cvterm_id'] = array('cv_id' => $cv_id);
  271. $statement = "sel_cvtermsynonym_sycv";
  272. }
  273. if ($cv_name) {
  274. $values['cvterm_id'] = array('cv_id' => array('name' => $cv_name));
  275. $statement = "sel_cvtermsynonym_sycv";
  276. }
  277. $options = array(
  278. 'statement_name' => $statement,
  279. 'case_insensitive_columns' => array('name')
  280. );
  281. $synonym = tripal_core_chado_select('cvtermsynonym', array('cvterm_id'), $values, $options);
  282. // if the synonym doens't exist or more than one record is returned then return false
  283. if(count($synonym) == 0) {
  284. return FALSE;
  285. }
  286. if(count($synonym) > 1) {
  287. return FALSE;
  288. }
  289. // get the cvterm
  290. $values = array('cvterm_id' => $synonym[0]->cvterm_id);
  291. $options = array('statement_name' => 'sel_cvterm_id');
  292. $cvterm = tripal_core_chado_select('cvterm', array('*'), $values, $options);
  293. return $cvterm[0];
  294. }
  295. /**
  296. * Create an options array to be used in a form element
  297. * which provides a list of all chado cvterms
  298. *
  299. * @param $cv_id
  300. * The chado cv_id;
  301. * only cvterms with the supplied cv_id will be returned
  302. * @return
  303. * An array(cvterm_id => name)
  304. * for each cvterm in the chado cvterm table where cv_id=that supplied
  305. *
  306. * @ingroup tripal_cv_api
  307. */
  308. function tripal_cv_get_cvterm_options($cv_id = 0) {
  309. if ($cv_id > 0) {
  310. $results = tripal_core_chado_select('cvterm', array('cvterm_id', 'name'), array('cv_id' => $cv_id));
  311. }
  312. else {
  313. $results = tripal_core_chado_select('cvterm', array('cvterm_id', 'name'), array());
  314. }
  315. $options = array();
  316. foreach ($results as $r) {
  317. $options[$r->cvterm_id] = $r->name;
  318. }
  319. return $options;
  320. }
  321. /**
  322. * Updates the cvtermpath table of Chado for the specified CV.
  323. *
  324. * @param $cv_id
  325. * The chado cv_id;
  326. * @param $job_id
  327. * This function is intended to be used with the Tripal Jobs API.
  328. * When this function is called as a job the $job_id is automatically
  329. * passed to this function.
  330. * @return
  331. * TRUE on success FALSE on failure
  332. *
  333. * @ingroup tripal_cv_api
  334. */
  335. function tripal_cv_update_cvtermpath($cvid, $job_id = NULL) {
  336. // TODO: need better error checking in this function
  337. // first get the controlled vocabulary name:
  338. $sql = "SELECT * FROM {cv} WHERE cv_id = :cv_id";
  339. $cv = chado_query($sql, array(':cv_id' => $cvid))->fetchObject();
  340. print "\nUpdating cvtermpath for $cv->name...\n";
  341. $previous = tripal_db_set_active('chado');
  342. try {
  343. $sql = "SELECT * FROM fill_cvtermpath(:name)";
  344. db_query($sql, array(':name' => $cv->name));
  345. tripal_db_set_active($previous);
  346. }
  347. catch (Exception $e) {
  348. tripal_db_set_active($previous);
  349. $error = $e->getMessage();
  350. watchdog('tripal_cv', "Could not fill cvtermpath table: @error", array('@error' => $error), WATCHDOG_ERROR);
  351. return FALSE;
  352. }
  353. return TRUE;
  354. }
  355. /**
  356. * Adds a controlled vocabular to the CV table of Chado.
  357. *
  358. * @param $name
  359. * The name of the controlled vocabulary. These are typically all lower case
  360. * with no special characters other than an undrescore (for spaces).
  361. * @param $comment
  362. * A description or definition of the vocabulary.
  363. *
  364. * @return
  365. * An object populated with fields from the newly added database.
  366. *
  367. * @ingroup tripal_cv_api
  368. */
  369. function tripal_cv_add_cv($name, $definition) {
  370. // insert/update values
  371. $ins_values = array(
  372. 'name' => $name,
  373. 'definition' => $definition
  374. );
  375. // see if the CV (default-namespace) exists already in the database
  376. $sel_values = array('name' => $name);
  377. $sel_options = array('statement_name' => 'sel_cv_na');
  378. $results = tripal_core_chado_select('cv', array('*'), $sel_values, $sel_options);
  379. // if it does not exists then add it
  380. if (count($results) == 0) {
  381. $ins_options = array('statement_name' => 'ins_cv_nade');
  382. $success = tripal_core_chado_insert('cv', $ins_values, $ins_options);
  383. if (!$success) {
  384. watchdog('tripal_cv', "Failed to create the CV record", NULL, WATCHDOG_WARNING);
  385. return FALSE;
  386. }
  387. $results = tripal_core_chado_select('cv', array('*'), $sel_values, $sel_options);
  388. }
  389. // if it already exists then do an update
  390. else {
  391. $upd_options = array('statement_name' => 'upd_cv_nade');
  392. $success = tripal_core_chado_update('cv', $sel_values, $ins_values, $upd_options);
  393. if (!$success) {
  394. watchdog('tripal_cv', "Failed to update the CV record", NULL, WATCHDOG_WARNING);
  395. return FALSE;
  396. }
  397. $results = tripal_core_chado_select('cv', array('*'), $sel_values, $sel_options);
  398. }
  399. // return the cv object
  400. return $results[0];
  401. }
  402. /**
  403. * Add's a CV term to the cvterm table. If the parent CV does not exist then
  404. * that too is added to the CV table. If the cvterm is a relationship term
  405. * then the $is_relationship argument should be set. The function will try
  406. * to first find the relationship in the relationship ontology for updating and
  407. * if it can't be found will add the relationship to the __global CV. All terms
  408. * must also have a corresponding database. This is specified in the term's
  409. * ID just before the colon (e.g. GO:003824). If the database does not exist
  410. * in the DB table then it will be added automatically. The accession (the
  411. * value just after the colon in the term's ID) will be added to the dbxref
  412. * table. If the CVterm already exists and $update is set (default) then the
  413. * cvterm is updated. If the CVTerm already exists and $update is not set, then
  414. * no changes are made and the CVTerm object is returned.
  415. *
  416. * @param $term
  417. * An associative array with the following keys: 'id', 'name' and 'namespace',
  418. * 'is_obsolete', and 'def'. Where 'id' is the term accession, 'name' is the
  419. * term name, 'namespace' is the CV name for the term, 'def' is the term
  420. * definition and 'is_obsolete' is present and set to 1 if the term is defunct.
  421. * The 'id' must be of the form <DB>:<ACCESSION>, where <DB> is the name of
  422. * the database to which the cvterm belongs and the <ACCESSION> is the
  423. * term's accession number in the database.
  424. * @param $defaultcv
  425. * Optional. The CV name to which the term
  426. * belongs. If this arugment is null or not provided then the function tries
  427. * to find a record in the CV table with the same name provided in the
  428. * $term[namespace]. If this field is provided then it overrides what the
  429. * value in $term[namespace]
  430. * @param $is_relationship
  431. * If this term is a relationship term then this value should be 1.
  432. * @param $update
  433. * By default this is set to 1. If the term exists it is automatically updated.
  434. * @param $dbname
  435. * In some cases the database name will not be part of the $term['id'] and it
  436. * needs to be explicitly set. Use this argument only if the database name
  437. * cannot be specififed in the term ID (e.g. <DB>:<ACCESSION>).
  438. *
  439. * @return
  440. * A CVTerm object
  441. *
  442. * @ingroup tripal_cv_api
  443. */
  444. function tripal_cv_add_cvterm($term, $defaultcv = '_global', $is_relationship = 0,
  445. $update = 1, $dbname = 'internal') {
  446. // get the term properties
  447. $id = $term['id'];
  448. $name = '';
  449. $cvname = '';
  450. $definition = '';
  451. $is_obsolete = 0;
  452. $accession = '';
  453. if (array_key_exists('name', $term)) {
  454. $name = $term['name'];
  455. }
  456. else {
  457. $name = $id;
  458. }
  459. if (array_key_exists('namespace', $term)) {
  460. $cvname = $term['namespace'];
  461. }
  462. else {
  463. $cvname = $defaultcv;
  464. }
  465. if (array_key_exists('def', $term)) {
  466. $definition = preg_replace('/^\"(.*)\"/', '\1', $term['def']);
  467. }
  468. else {
  469. $definition = '';
  470. }
  471. if (array_key_exists('is_obsolete', $term)) {
  472. $is_obsolete = $term['is_obsolete'];
  473. if (strcmp($is_obsolete, 'true') == 0) {
  474. $is_obsolete = 1;
  475. }
  476. }
  477. if (!$name and !$id) {
  478. watchdog('tripal_cv', "Cannot find cvterm without 'id' or 'name'", NULL, WATCHDOG_WARNING);
  479. return 0;
  480. }
  481. if (!$id) {
  482. $id = $name;
  483. }
  484. // get the accession and the database from the cvterm id
  485. if ($dbname) {
  486. $accession = $id;
  487. }
  488. if (preg_match('/^.+?:.*$/', $id)) {
  489. $accession = preg_replace('/^.+?:(.*)$/', '\1', $id);
  490. $dbname = preg_replace('/^(.+?):.*$/', '\1', $id);
  491. }
  492. // check that we have a database name, give a different message if it's a relationship
  493. if ($is_relationship and !$dbname) {
  494. watchdog('tripal_cv', "A database name is not provided for this relationship term: $id", NULL, WATCHDOG_WARNING);
  495. return 0;
  496. }
  497. if (!$is_relationship and !$dbname) {
  498. watchdog('tripal_cv', "A database identifier is missing from the term: $id", NULL, WATCHDOG_WARNING);
  499. return 0;
  500. }
  501. // make sure the CV name exists
  502. $cv = tripal_cv_add_cv($cvname, '');
  503. if (!$cv) {
  504. watchdog('tripal_cv', "Cannot find namespace '$cvname' when adding/updating $id", NULL, WATCHDOG_WARNING);
  505. return 0;
  506. }
  507. // this SQL statement will be used a lot to find a cvterm so just set it
  508. // here for easy reference below. Because CV terms can change their names
  509. // but accessions don't change, the following SQL finds cvterms based on
  510. // their accession rather than the name
  511. $cvtermsql = "
  512. SELECT CVT.name, CVT.cvterm_id, CV.cv_id, CV.name as cvname,
  513. DB.name as dbname, DB.db_id, DBX.accession
  514. FROM {cvterm} CVT
  515. INNER JOIN {dbxref} DBX on CVT.dbxref_id = DBX.dbxref_id
  516. INNER JOIN {db} DB on DBX.db_id = DB.db_id
  517. INNER JOIN {cv} CV on CV.cv_id = CVT.cv_id
  518. WHERE DBX.accession = :accession and DB.name = :name
  519. ";
  520. // add the database. The function will just return the DB object if the
  521. // database already exists.
  522. $db = tripal_db_add_db($dbname);
  523. if (!$db) {
  524. watchdog('tripal_cv', "Cannot find database '$dbname' in Chado.", NULL, WATCHDOG_WARNING);
  525. return 0;
  526. }
  527. // the cvterm table has two unique dependencies. We need to check both.
  528. // first check the (name, cv_id, is_obsolete) constraint
  529. $values = array(
  530. 'name' => $name,
  531. 'is_obsolete' => $is_obsolete,
  532. 'cv_id' => array(
  533. 'name' => $cvname,
  534. ),
  535. );
  536. $options = array('statement_name' => 'sel_cvterm_c1');
  537. $result = tripal_core_chado_select('cvterm', array('*'), $values, $options);
  538. // if the constraint is met then let's check it to see if
  539. // the database name matches the one we have been provided
  540. if (count($result) == 1) {
  541. $cvterm = $result[0];
  542. // get the dbxref record
  543. $values = array('dbxref_id' => $cvterm->dbxref_id);
  544. $options = array('statement_name' => 'sel_dbxref_id');
  545. $result = tripal_core_chado_select('dbxref', array('*'), $values, $options);
  546. $dbxref = $result[0];
  547. // get the db
  548. $values = array('db_id' => $dbxref->db_id);
  549. $options = array('statement_name' => 'sel_db_id');
  550. $result = tripal_core_chado_select('db', array('*'), $values, $options);
  551. $db_check = $result[0];
  552. // the database name for this existing term does not match that of the
  553. // one provided to this function. The CV name matches otherwise we
  554. // wouldn't have made it this far. So, let's swap the database for
  555. // this term
  556. if ($db_check->name != $db->name) {
  557. // look to see if the correct dbxref record already exists for this database
  558. $values = array(
  559. 'db_id' => $db->db_id,
  560. 'accession' => $accession,
  561. );
  562. $options = array('statement_name' => 'sel_dbxref_idac');
  563. $result = tripal_core_chado_select('dbxref', array('*'), $values, $options);
  564. // if we already have a good dbxref then we want to update our cvterm
  565. // to use this dbxref
  566. if (count($result) > 0) {
  567. $dbxref = $result[0];
  568. $match = array('cvterm_id' => $cvterm->cvterm_id);
  569. $values = array('dbxref_id' => $dbxref->dbxref_id);
  570. $options = array('statement_name' => 'upd_cvterm_db');
  571. $success = tripal_core_chado_update('cvterm', $match, $values, $options);
  572. if (!$success) {
  573. watchdog('tripal_cv', "Failed to correct the dbxref id for the cvterm " .
  574. "'$name' (id: $accession), for database $dbname", NULL, WATCHDOG_WARNING);
  575. return 0;
  576. }
  577. }
  578. // if we don't have the record then we want to delete our cvterm and let the code
  579. // below recreate it with the correct info
  580. else {
  581. $match = array('cvterm_id' => $cvterm->cvterm_id);
  582. $options = array('statement_name' => 'del_cvterm_cv');
  583. tripal_core_chado_delete('cvterm', $match, $options);
  584. }
  585. }
  586. // check that the accession matches. Sometimes an OBO can define the same term
  587. // multiple times but with different accessions. If this is the case we
  588. // can't do an insert or it will violate the constraint in the cvterm table.
  589. // so we'll need to add the record to the cvterm_dbxref table instead
  590. if ($dbxref->accession != $accession) {
  591. // get/add the dbxref fort his term
  592. $dbxref_new = tripal_db_add_dbxref($db->db_id, $accession);
  593. if (!$dbxref_new) {
  594. watchdog('tripal_cv', "Failed to find or insert the dbxref record for cvterm, " .
  595. "$name (id: $accession), for database $dbname", NULL, WATCHDOG_WARNING);
  596. return 0;
  597. }
  598. // check to see if the cvterm_dbxref record already exists
  599. $values = array(
  600. 'cvterm_id' => $cvterm->cvterm_id,
  601. 'dbxref_id' => $dbxref_new->dbxref_id,
  602. 'is_for_definition' => 1,
  603. );
  604. $options = array('statement_name' => 'sel_cvtermdbxref_cvdbis');
  605. $result = tripal_core_chado_select('cvterm_dbxref', array('*'), $values, $options);
  606. // if the cvterm_dbxref record does not exists then add it
  607. if (count($result)==0) {
  608. $options = array(
  609. 'statement_name' => 'ins_cvtermdbxref_cvdbis',
  610. 'return_record' => FALSE,
  611. );
  612. $success = tripal_core_chado_insert('cvterm_dbxref', $values, $options);
  613. if (!$success) {
  614. watchdog('tripal_cv', "Failed to find or insert the cvterm_dbxref record for a " .
  615. "duplicated cvterm: $name (id: $accession), for database $dbname", NULL, WATCHDOG_WARNING);
  616. return 0;
  617. }
  618. }
  619. // get the original cvterm with the same name and return that.
  620. $result = chado_query($cvtermsql, array(':accession' => $dbxref->accession, ':name' => $dbname));
  621. $cvterm = $result->fetchObject();
  622. return $cvterm;
  623. }
  624. // continue on, we've fixed the record if the db_id did not match,
  625. // we can now perform and updated if we need to.
  626. }
  627. // get the CVterm record
  628. $result = chado_query($cvtermsql, array(':accession' => $accession, ':name' => $dbname));
  629. $cvterm = $result->fetchObject();
  630. if (!$cvterm) {
  631. // check to see if the dbxref exists if not, add it
  632. $dbxref = tripal_db_add_dbxref($db->db_id, $accession);
  633. if (!$dbxref) {
  634. watchdog('tripal_cv', "Failed to find or insert the dbxref record for cvterm, " .
  635. "$name (id: $accession), for database $dbname", NULL, WATCHDOG_WARNING);
  636. return 0;
  637. }
  638. // check to see if the dbxref already has an entry in the cvterm table
  639. // this is the second constraint in the cvterm table
  640. $values = array('dbxref_id' => $dbxref->dbxref_id);
  641. $options = array('statement_name' => 'sel_cvterm_db');
  642. $check = tripal_core_chado_select('cvterm', array('cvterm_id'), $values, $options);
  643. if (count($check) == 0) {
  644. // now add the cvterm
  645. $ins_values = array(
  646. 'cv_id' => $cv->cv_id,
  647. 'name' => $name,
  648. 'definition' => $definition,
  649. 'dbxref_id' => $dbxref->dbxref_id,
  650. 'is_obsolete' => $is_obsolete,
  651. 'is_relationshiptype' => $is_relationship,
  652. );
  653. $ins_options = array('statement_name' => 'ins_cvterm_all');
  654. $success = tripal_core_chado_insert('cvterm', $ins_values, $ins_options);
  655. if (!$success) {
  656. if (!$is_relationship) {
  657. watchdog('tripal_cv', "Failed to insert the term: $name ($dbname)", NULL, WATCHDOG_WARNING);
  658. return 0;
  659. }
  660. else {
  661. watchdog('tripal_cv', "Failed to insert the relationship term: $name (cv: " . $cvname . " db: $dbname)", NULL, WATCHDOG_WARNING);
  662. return 0;
  663. }
  664. }
  665. }
  666. // this dbxref already exists in the cvterm table
  667. else {
  668. watchdog('tripal_cv', "The dbxref already exists for another cvterm record: $name (cv: " . $cvname . " db: $dbname)", NULL, WATCHDOG_WARNING);
  669. return 0;
  670. }
  671. $result = chado_query($cvtermsql, array(':accession' => $accession, ':name' => $dbname));
  672. $cvterm = $result->fetchObject();
  673. }
  674. // upate the cvterm
  675. elseif ($update) {
  676. $match = array('cvterm_id' => $cvterm->cvterm_id);
  677. $upd_values = array(
  678. 'name' => $name,
  679. 'definition' => $definition,
  680. 'is_obsolete' => $is_obsolete,
  681. 'is_relationshiptype' => $is_relationship,
  682. );
  683. $upd_options = array('statement_name' => 'upd_cvterm_nadeisis');
  684. $success = tripal_core_chado_update('cvterm', $match, $upd_values, $upd_options);
  685. if (!$success) {
  686. watchdog('tripal_cv', "Failed to update the term: $name", NULL, WATCHDOG_WARNING);
  687. return 0;
  688. }
  689. $result = chado_query($cvtermsql, array(':accession' => $accession, ':name' => $dbname));
  690. $cvterm = $result->fetchObject();
  691. }
  692. else {
  693. // do nothing, we have the cvterm but we don't want to update
  694. }
  695. // return the cvterm
  696. return $cvterm;
  697. }
  698. /**
  699. * This function defines the custom tables that will be created
  700. * in the chado schema.
  701. *
  702. * @ingroup tripal_cv_api
  703. */
  704. function tripal_cv_get_custom_tables($table = NULL) {
  705. if (!$table or strcmp($table, 'tripal_obo_temp')==0) {
  706. $schema['tripal_obo_temp'] = array(
  707. 'table' => 'tripal_obo_temp',
  708. 'fields' => array(
  709. 'id' => array(
  710. 'type' => 'varchar',
  711. 'length' => '255',
  712. 'not null' => TRUE,
  713. ),
  714. 'stanza' => array(
  715. 'type' => 'text',
  716. 'not null' => TRUE,
  717. ),
  718. 'type' => array(
  719. 'type' => 'varchar',
  720. 'length' => '50',
  721. 'not null' => TRUE,
  722. ),
  723. ),
  724. 'indexes' => array(
  725. 'tripal_obo_temp_idx0' => array('id'),
  726. 'tripal_obo_temp_idx0' => array('type'),
  727. ),
  728. 'unique keys' => array(
  729. 'tripal_obo_temp_uq0' => array('id'),
  730. ),
  731. );
  732. }
  733. return $schema;
  734. }
  735. /**
  736. * This function allows other modules to programatically
  737. * submit an ontology for loading into Chado. This function
  738. * will add a job to the Jobs subsystem for parsing the ontology.
  739. * You can either pass a known OBO ID to the function or the URL
  740. * or full path the the ontology file. If a URL or file name is
  741. * passed then the $obo_name argument must also be provided. If
  742. * this is the first time the ontology has been provided to Tripal
  743. * then it will be added to the database and will be assigned a
  744. * unique OBO ID.
  745. *
  746. * @param $obo_id
  747. * If the ontology is already loaded into the Tripal tables then
  748. * use this argument to specify the unique ID for the ontology
  749. * that will be loaded
  750. * @param $obo_name
  751. * If the OBO has not been added before then use this argument
  752. * to specify the human readable name of the ontology.
  753. * @param $obo_url
  754. * If the OBO to be loaded is located on a remote server then
  755. * use this argument to provide the URL.
  756. * @param $obo_file
  757. * If the OBO is housed on the local file system of the server then
  758. * use this argument to specify the full path.
  759. *
  760. * @return
  761. * returns the job_id of the submitted job or FALSE if the job was not added
  762. *
  763. * @ingroup tripal_cv_api
  764. */
  765. function tripal_cv_submit_obo_job($obo_id = NULL, $obo_name = NULL, $obo_url = NULL, $obo_file = NULL) {
  766. global $user;
  767. if ($obo_id) {
  768. $sql = "SELECT * FROM {tripal_cv_obo} WHERE obo_id = :obo_id";
  769. $obo = db_query($sql, array(':obo_id' => $obo_id))->fetchObject();
  770. $args = array($obo_id);
  771. return tripal_add_job("Load OBO $obo->name", 'tripal_cv',
  772. "tripal_cv_load_obo_v1_2_id", $args, $user->uid);
  773. }
  774. else {
  775. if ($obo_url) {
  776. $args = array($obo_name, $obo_url);
  777. return tripal_add_job("Load OBO $obo_name", 'tripal_cv',
  778. "tripal_cv_load_obo_v1_2_url", $args, $user->uid);
  779. }
  780. elseif ($obo_file) {
  781. $args = array($obo_name, $obo_file);
  782. return tripal_add_job("Load OBO $obo_name", 'tripal_cv',
  783. "tripal_cv_load_obo_v1_2_file", $args, $user->uid);
  784. }
  785. }
  786. return FALSE;
  787. }
  788. /**
  789. * Add the obo to the tripal_cv_obo table in the Drupal database
  790. *
  791. * @param $name
  792. * The human readable name of this ontology
  793. * @param $path
  794. * The file path or URL of the ontology
  795. *
  796. * @return
  797. * Returns the ontology ID
  798. *
  799. * @ingroup tripal_cv_api
  800. */
  801. function tripal_cv_add_obo_ref($name, $path) {
  802. $record = new stdClass;
  803. $record->name = $name;
  804. $record->path = $path;
  805. drupal_write_record('tripal_cv_obo', $record);
  806. return $record->obo_id;
  807. }