1: <?php
  2: 
  3: namespace cli_import;
  4: 
  5: require_once ROOT . 'classes/AbstractImporter.php';
  6: 
  7: /**
  8:  * importer for differential expressions
  9:  */
 10: class Importer_Differential_Expressions extends AbstractImporter {
 11: 
 12:     /**
 13:      * use for array_walk. converts 'NA', 'Inf' and '-Inf' to their postgres counterparts
 14:      * @param in:string,out:float $value will be changed in-place
 15:      * @param string $key neccessary for array_walk. will not be used
 16:      */
 17:     static function convertDbl(&$value, $key) {
 18:         if ($value == '-Inf')
 19:             $value = '-Infinity';
 20:         else if ($value == 'Inf')
 21:             $value = 'Infinity';
 22:         else if ($value == 'NA')
 23:             $value = 'NaN';
 24:         else if (floatval($value) > 0 && floatval($value) < 1e-307) {
 25:             $value = 0;
 26:         }
 27:     }
 28: 
 29:     /**
 30:      * @inheritDoc
 31:      */
 32:     static function import($options) {
 33: 
 34:         $filename = $options['file'];
 35:         $analysis_id = $options['analysis_id'];
 36:         $biomaterial_parentA_name = $options['conditionGroupA'];
 37:         $biomaterial_parentB_name = $options['conditionGroupB'];
 38: 
 39:         $lines_total = trim(`wc -l $filename | cut -d' ' -f1`);
 40:         self::setLineCount($lines_total);
 41: 
 42:         global $db;
 43:         $lines_imported = 0;
 44: 
 45:         $lines_skipped = 0;
 46: #IDE type hint
 47:         if (false)
 48:             $db = new PDO ();
 49: 
 50:         try {
 51:             $db->beginTransaction();
 52: 
 53:             //get biomaterial A id
 54:             $statement_get_biomaterial_id = $db->prepare('SELECT b.biomaterial_id, bp.value AS type FROM biomaterial b JOIN biomaterialprop bp ON (b.biomaterial_id = bp.biomaterial_id) WHERE b.name=:name AND bp.type_id = ' . CV_BIOMATERIAL_TYPE . ' LIMIT 1');
 55:             $statement_get_biomaterial_id->bindValue('name', $biomaterial_parentA_name);
 56:             $statement_get_biomaterial_id->execute();
 57:             $rowa = $statement_get_biomaterial_id->fetch(\PDO::FETCH_ASSOC);
 58:             if ($statement_get_biomaterial_id->rowCount() == 0) {
 59:                 throw new ErrorException(sprintf('Biomaterial with this name not defined (%s)', $biomaterial_parentA_name));
 60:             }
 61:             //is it a condition?
 62:             if ($rowa['type'] != 'condition') {
 63:                 throw new ErrorException(sprintf('This biomaterial is not of type condition! (%s)', $biomaterial_parentA_name));
 64:             }
 65:             $biomaterial_parentA_id = $rowa['biomaterial_id'];
 66: 
 67:             //get biomaterial B id
 68:             $statement_get_biomaterial_id->bindValue('name', $biomaterial_parentB_name);
 69:             $statement_get_biomaterial_id->execute();
 70:             $rowb = $statement_get_biomaterial_id->fetch(\PDO::FETCH_ASSOC);
 71:             if ($statement_get_biomaterial_id->rowCount() == 0) {
 72:                 throw new ErrorException(sprintf('Biomaterial with this name not defined (%s)', $biomaterial_parentB_name));
 73:             }
 74:             //is it a condition?
 75:             if ($rowb['type'] != 'condition') {
 76:                 throw new ErrorException(sprintf('This biomaterial is not of type condition! (%s)', $biomaterial_parentB_name));
 77:             }
 78:             $biomaterial_parentB_id = $rowb['biomaterial_id'];
 79: 
 80: 
 81:             //does A have samples?
 82:             $statement_test_biomaterial_children = $db->prepare('SELECT biomaterial_relationship_id FROM biomaterial_relationship WHERE object_id=:parent LIMIT 1');
 83:             $statement_test_biomaterial_children->bindValue('parent', $biomaterial_parentA_id);
 84:             $statement_test_biomaterial_children->execute();
 85:             if (!($statement_test_biomaterial_children->fetchColumn())) {
 86:                 throw new ErrorException(sprintf('Biomaterial has no children (%s)', $biomaterial_parentA_name));
 87:             }
 88: 
 89:             //does B have samples?
 90:             $statement_test_biomaterial_children->bindValue('parent', $biomaterial_parentB_id);
 91:             $statement_test_biomaterial_children->execute();
 92:             if (!($statement_test_biomaterial_children->fetchColumn())) {
 93:                 throw new ErrorException(sprintf('Biomaterial has no children (%s)', $biomaterial_parentB_name));
 94:             }
 95: 
 96: 
 97:             $dummy = null;
 98:             $feature_name = null;
 99:             $param_baseMean = null;
100:             $param_baseMeanA = null;
101:             $param_baseMeanB = null;
102:             $param_foldChange = null;
103:             $param_log2foldChange = null;
104:             $param_pval = null;
105:             $param_pvaladj = null;
106:             $param_feature_uniquename = null;
107: 
108:             //query for insertion of diffexp
109:             $query_insert_diffexp = <<<EOF
110: INSERT INTO diffexpresult(analysis_id, feature_id, biomateriala_id, biomaterialb_id, baseMean, baseMeanA, baseMeanB, foldChange, log2foldChange, pval, pvaladj)
111: SELECT :analysis_id, feature_id, :biomaterialA_id, :biomaterialB_id, :baseMean, :baseMeanA, :baseMeanB, :foldChange, :log2foldChange, :pval, :pvaladj
112: FROM feature WHERE uniquename = :feature_uniquename AND organism_id = :organism
113: EOF;
114: 
115:             $statement_insert_diffexp = $db->prepare($query_insert_diffexp);
116:             $statement_insert_diffexp->bindValue('analysis_id', $analysis_id, PDO::PARAM_INT);
117:             $statement_insert_diffexp->bindValue('biomaterialA_id', $biomaterial_parentA_id, PDO::PARAM_INT);
118:             $statement_insert_diffexp->bindValue('biomaterialB_id', $biomaterial_parentB_id, PDO::PARAM_INT);
119:             $statement_insert_diffexp->bindParam('baseMean', $param_baseMean, PDO::PARAM_STR);
120:             $statement_insert_diffexp->bindParam('baseMeanA', $param_baseMeanA, PDO::PARAM_STR);
121:             $statement_insert_diffexp->bindParam('baseMeanB', $param_baseMeanB, PDO::PARAM_STR);
122:             $statement_insert_diffexp->bindParam('foldChange', $param_foldChange, PDO::PARAM_STR);
123:             $statement_insert_diffexp->bindParam('log2foldChange', $param_log2foldChange, PDO::PARAM_STR);
124:             $statement_insert_diffexp->bindParam('pval', $param_pval, PDO::PARAM_STR);
125:             $statement_insert_diffexp->bindParam('pvaladj', $param_pvaladj, PDO::PARAM_STR);
126: 
127:             $statement_insert_diffexp->bindParam('feature_uniquename', $param_feature_uniquename, PDO::PARAM_STR);
128:             $statement_insert_diffexp->bindValue('organism', DB_ORGANISM_ID, PDO::PARAM_INT);
129: 
130: 
131:             $file = fopen($filename, 'r');
132:             if (feof($file))
133:                 return;
134: #just skipping header
135:             fgets($file);
136: 
137:             while (($line = fgetcsv($file, 0, ",")) !== false) {
138:                 array_walk($line, array('Importer_Differential_Expressions', 'convertDbl'));
139:                 list($dummy, $feature_name, $param_baseMean, $param_baseMeanA, $param_baseMeanB, $param_foldChange, $param_log2foldChange, $param_pval, $param_pvaladj) = $line;
140:                 //this importer may encounter lines of NA NA NA NA (...batman!). skip these
141:                 if ($feature_name == 'NaN') {
142:                     $lines_skipped++;
143:                     continue;
144:                 }
145: 
146:                 //insert diffexp
147:                 $param_feature_uniquename = IMPORT_PREFIX . "_" . $feature_name;
148:                 $statement_insert_diffexp->execute();
149: 
150:                 $lines_feature_skipped += ($statement_insert_diffexp->rowCount() == 0) ? 1 : 0;
151:                 self::updateProgress(++$lines_imported);
152:             }
153:             self::preCommitMsg();
154:             if (!$db->commit()) {
155:                 $err = $db->errorInfo();
156:                 throw new ErrorException($err[2], ERRCODE_TRANSACTION_NOT_COMPLETED, 1);
157:             }
158:         } catch (\Exception $error) {
159:             $db->rollback();
160:             throw $error;
161:         }
162:         return array(LINES_IMPORTED => $lines_imported, 'lines_featurenotfound_skipped' => $lines_feature_skipped, 'lines_NA_skipped' => $lines_skipped);
163:     }
164: 
165:     /**
166:      * @inheritDoc
167:      */
168:     public static function CLI_getCommand(\Console_CommandLine $parser) {
169:         $command = parent::CLI_getCommand($parser);
170: 
171:         $command->addOption('analysis_id', array(
172:             'short_name' => '-a',
173:             'long_name' => '--analysis_id',
174:             'description' => 'analysis id'
175:         ));
176:         $command->addOption('conditionGroupA', array(
177:             'short_name' => '-A',
178:             'long_name' => '--conditionA',
179:             'description' => 'condition A name'
180:         ));
181:         $command->addOption('conditionGroupB', array(
182:             'short_name' => '-B',
183:             'long_name' => '--conditionB',
184:             'description' => 'condition B name'
185:         ));
186:         return $command;
187:     }
188: 
189:     /**
190:      * @inheritDoc
191:      */
192:     public static function CLI_checkRequiredOpts(\Console_CommandLine_Result $command) {
193:         parent::CLI_checkRequiredOpts($command);
194:         $options = $command->options;
195:         AbstractImporter::dieOnMissingArg($options, 'analysis_id');
196:         AbstractImporter::dieOnMissingArg($options, 'conditionGroupA');
197:         AbstractImporter::dieOnMissingArg($options, 'conditionGroupB');
198:     }
199: 
200:     /**
201:      * @inheritDoc
202:      */
203:     public static function CLI_commandDescription() {
204:         return "Importer for differential expression results";
205:     }
206: 
207:     /**
208:      * @inheritDoc
209:      */
210:     public static function CLI_commandName() {
211:         return 'differential_expressions';
212:     }
213: 
214:     /**
215:      * @inheritDoc
216:      */
217:     public static function CLI_longHelp() {
218:         return <<<EOF
219: 
220: File Format looks like this (\033[0;31mFirst line will be skipped\033[0m):
221: ,id,baseMean,baseMeanA,baseMeanB,foldChange,log2FoldChange,pval,padj
222: 6,comp230079_c0,249.687338527051,206.660251316251,292.714425737851,1.41640409257952,0.502232917392478,2.32555262831702e-08,9.65409100672613e-08
223: 9,comp234683_c0,1904.88401956508,1811.60920428892,1998.15883484125,1.10297454335664,0.141399493923902,0.000466092095479145,0.00137346251047776
224:    
225: \033[0;31mThis import requires a successful Sequence ID Import!\033[0m
226: EOF;
227:     }
228: 
229: }
230: 
231: ?>
232: