1: <?php
  2: 
  3: namespace cli_import;
  4: 
  5: require_once ROOT . 'classes/AbstractImporter.php';
  6: 
  7: /**
  8:  * importer for sequence ids and mapping
  9:  */
 10: class Importer_Sequence_Ids extends AbstractImporter {
 11: 
 12:     /**
 13:      * get import prefix id, insert into dbxref if neccessary
 14:      * @global \PDO $db
 15:      * @global String DB_NAME_IMPORTS
 16:      * @global String IMPORT_PREFIX
 17:      * @return int
 18:      */
 19:     public static function get_import_dbxref() {
 20:         global $db;
 21:         $stm_get_import_prefix_id = $db->prepare('SELECT get_or_insert_dbxref(?, ?)');
 22:         $stm_get_import_prefix_id->execute(array(DB_NAME_IMPORTS, IMPORT_PREFIX));
 23:         $import_prefix_id = $stm_get_import_prefix_id->fetchColumn();
 24:         unset($stm_get_import_prefix_id);
 25: 
 26:         return $import_prefix_id;
 27:     }
 28: 
 29:     /**
 30:      * @inheritDoc
 31:      */
 32:     static function import($options) {
 33:         $filename = $options['file'];
 34:         $file_type = $options['file_type'];
 35: 
 36:         $lines_total = trim(`wc -l $filename | cut -d' ' -f1`);
 37:         self::setLineCount($lines_total);
 38: 
 39:         global $db;
 40:         if (false)
 41:             $db = new PDO();
 42: 
 43:         $lines_imported = 0;
 44:         $isoforms_added = 0;
 45:         $unigenes_added = 0;
 46: 
 47:         #pre-initialize variables to bind statement parameters
 48:         $param_unigene_name = null;
 49:         $param_unigene_uniq = null;
 50:         $param_isoform_name = null;
 51:         $param_isoform_uniq = null;
 52:         $param_unigene_lastid = null;
 53: 
 54:         try {
 55:             $db->beginTransaction();
 56:             $import_prefix_id = self::get_import_dbxref();
 57: 
 58:             # link import dbxref to organism if not already linked
 59:             $stm_link_organism_import_dbxref = $db->prepare("SELECT * FROM organism_dbxref WHERE organism_id=? AND dbxref_id=?");
 60:             $stm_link_organism_import_dbxref->execute(array(DB_ORGANISM_ID, $import_prefix_id));
 61:             if ($stm_link_organism_import_dbxref->rowCount() == 0)
 62:                 $db->prepare("INSERT INTO organism_dbxref (organism_id, dbxref_id) VALUES (?, ?)")->execute(array(DB_ORGANISM_ID, $import_prefix_id));
 63:             unset($stm_link_organism_import_dbxref);
 64: 
 65:             #get unigene id for existing unigene
 66:             $stm_sel_unigene_byUniquename = $db->prepare('SELECT feature_id FROM feature WHERE uniquename=:uniquename AND organism_id=:organism_id');
 67:             $stm_sel_unigene_byUniquename->bindValue('organism_id', DB_ORGANISM_ID, PDO::PARAM_INT);
 68:             $stm_sel_unigene_byUniquename->bindParam('uniquename', $param_unigene_uniq);
 69: 
 70: 
 71:             # insert unigene
 72:             $stm_ins_unigene = $db->prepare('INSERT INTO feature (name, uniquename, type_id, organism_id, dbxref_id) VALUES (:name, :uniquename, :type_id, :organism_id, :dbxref_id) RETURNING feature_id');
 73:             $stm_ins_unigene->bindValue('type_id', CV_UNIGENE, PDO::PARAM_INT);
 74:             $stm_ins_unigene->bindValue('organism_id', DB_ORGANISM_ID, PDO::PARAM_INT);
 75:             $stm_ins_unigene->bindParam('name', $param_unigene_name, PDO::PARAM_STR);
 76:             $stm_ins_unigene->bindParam('uniquename', $param_unigene_uniq);
 77:             $stm_ins_unigene->bindValue('dbxref_id', $import_prefix_id, PDO::PARAM_INT);
 78: 
 79:             #insert feature
 80:             $stm_ins_isoform = $db->prepare('INSERT INTO feature (name, uniquename, type_id, organism_id, dbxref_id) VALUES (:name, :uniquename, :type_id, :organism_id, :dbxref_id)');
 81:             $stm_ins_isoform->bindValue('type_id', CV_ISOFORM, PDO::PARAM_INT);
 82:             $stm_ins_isoform->bindValue('organism_id', DB_ORGANISM_ID, PDO::PARAM_INT);
 83:             $stm_ins_isoform->bindParam('name', $param_isoform_name, PDO::PARAM_STR);
 84:             $stm_ins_isoform->bindParam('uniquename', $param_isoform_uniq, PDO::PARAM_STR);
 85:             $stm_ins_isoform->bindValue('dbxref_id', $import_prefix_id, PDO::PARAM_INT);
 86: 
 87:             #link isoform to unigene
 88:             $stm_ins_feature_rel = $db->prepare('INSERT INTO feature_relationship (subject_id, type_id, object_id) VALUES (currval(\'feature_feature_id_seq\'), :type_id, :parent)');
 89:             $stm_ins_feature_rel->bindValue('type_id', CV_RELATIONSHIP_UNIGENE_ISOFORM, PDO::PARAM_INT);
 90:             $stm_ins_feature_rel->bindParam('parent', $param_unigene_lastid, PDO::PARAM_INT);
 91: 
 92: 
 93:             $file = fopen($filename, 'r');
 94: 
 95:             $last_unigene = "";
 96:             while (($line = fgetcsv($file, 0, "\t")) !== false) {
 97:                 switch ($file_type) {
 98:                     case 'only_isoforms':
 99:                         $param_isoform_name = $line[0];
100:                         $param_isoform_uniq = IMPORT_PREFIX . "_" . $param_isoform_name;
101:                         $stm_ins_isoform->execute();
102:                         $isoforms_added++;
103:                         break;
104: 
105:                     case 'only_unigenes':
106:                         $param_unigene_name = $line[0];
107:                         $param_unigene_uniq = IMPORT_PREFIX . "_" . $param_unigene_name;
108:                         $stm_ins_unigene->execute();
109:                         $unigenes_added++;
110:                         break;
111: 
112:                     case 'map':
113:                         #split into parts
114:                         list($param_unigene_name, $param_isoform_name) = $line;
115: 
116:                         #this lines unigene is not the cached unigene from last line.
117:                         if ($last_unigene != $param_unigene_name) {
118:                             $param_unigene_uniq = IMPORT_PREFIX . "_" . $param_unigene_name;
119:                             #check if unigene exists
120:                             $stm_sel_unigene_byUniquename->execute();
121:                             $param_unigene_lastid = $stm_sel_unigene_byUniquename->fetchColumn();
122:                             #if it does not exist, insert it
123:                             if (!$param_unigene_lastid) {
124:                                 $stm_ins_unigene->execute();
125:                                 $unigenes_added++;
126:                                 # get last insert id (see query: 'RETURNING feature_id'), set id for feature_relationship insert
127:                                 $param_unigene_lastid = $stm_ins_unigene->fetchColumn();
128:                             }
129:                             # set for test to skip this unigene in the future
130:                             $last_unigene = $param_unigene_name;
131:                         }
132: 
133:                         if (!empty($param_isoform_name)) {
134:                             # set last value, execute insert
135:                             $param_isoform_uniq = IMPORT_PREFIX . "_" . $param_isoform_name;
136:                             $stm_ins_isoform->execute();
137:                             $isoforms_added++;
138: 
139:                             # insert feature_relationship
140:                             $stm_ins_feature_rel->execute();
141:                         }
142:                         break;
143:                 }
144: 
145:                 self::updateProgress(++$lines_imported);
146:             }
147:             self::preCommitMsg();
148:             if (!$db->commit()) {
149:                 $err = $db->errorInfo();
150:                 throw new ErrorException($err[2], ERRCODE_TRANSACTION_NOT_COMPLETED, 1);
151:             }
152:         } catch (\Exception $error) {
153:             $db->rollback();
154:             throw $error;
155:         }
156:         return array(LINES_IMPORTED => $lines_imported, 'unigenes_added' => $unigenes_added, 'isoforms_added' => $isoforms_added);
157:     }
158: 
159:     /**
160:      * @inheritDoc
161:      */
162:     public static function CLI_getCommand(\Console_CommandLine $parser) {
163:         $command = parent::CLI_getCommand($parser);
164:         $command->addOption('file_type', array(
165:             'short_name' => '-t',
166:             'long_name' => '--file_type',
167:             'description' => "file format description. one of ('map', 'only_isoforms', 'only_unigenes'). defaults to 'map'",
168:             'choices' => array('map', 'only_isoforms', 'only_unigenes'),
169:             'default' => 'map'
170:         ));
171:         return $command;
172:     }
173: 
174:     /**
175:      * @inheritDoc
176:      */
177:     public static function CLI_commandName() {
178:         return 'sequence_ids';
179:     }
180: 
181:     /**
182:      * @inheritDoc
183:      */
184:     public static function CLI_commandDescription() {
185:         return "Sequence ID Importer";
186:     }
187: 
188:     /**
189:      * @inheritDoc
190:      */
191:     public static function CLI_longHelp() {
192:         return <<<EOF
193: 
194: This command creates the database features all other commands work with.
195: You can either import a line-separated file with just the sequence ids you use throughout the files that will be imported or you can import a map, which maps isoforms on unigenes.
196:    
197: File format "only_isoforms":
198: isoform1
199: isoform2
200: isoform3
201: 
202: File format "only_unigenes":
203: unigene1
204: unigene2
205: unigene3
206: 
207: File Format "map" (tab-separated):
208: unigene1    isoform1
209: unigene1    isoform2
210: unigene2    isoform3
211: unigene3    isoform4
212: EOF;
213:     }
214: 
215: }
216: 
217: ?>
218: