|
|
@@ -1,26 +1,26 @@
|
|
|
<?php
|
|
|
-
|
|
|
/**
|
|
|
* @file
|
|
|
- * @todo Add file header description
|
|
|
+ * Provides fasta loading functionality. Creates features based on their specification
|
|
|
+ * in a fasta file.
|
|
|
*/
|
|
|
|
|
|
-
|
|
|
/**
|
|
|
* @defgroup fasta_loader FASTA Feature Loader
|
|
|
* @ingroup tripal_feature
|
|
|
* @{
|
|
|
- * Provides fasta loading functionality. Creates features based on their specification in a fasta file.
|
|
|
+ * Provides fasta loading functionality. Creates features based on their specification
|
|
|
+ * in a fasta file.
|
|
|
* @}
|
|
|
- *
|
|
|
+ *
|
|
|
*/
|
|
|
|
|
|
/**
|
|
|
- *
|
|
|
+ * The form to submit a fasta loading job
|
|
|
*
|
|
|
* @ingroup fasta_loader
|
|
|
*/
|
|
|
-function tripal_feature_fasta_load_form( ) {
|
|
|
+function tripal_feature_fasta_load_form() {
|
|
|
|
|
|
$form['fasta_file']= array(
|
|
|
'#type' => 'textfield',
|
|
|
@@ -100,7 +100,7 @@ function tripal_feature_fasta_load_form( ) {
|
|
|
t('Name'),
|
|
|
t('Unique name'),
|
|
|
),
|
|
|
- '#description' => t('Used for "updates only" or "insert and update" methods. Not required if method type is "insert".
|
|
|
+ '#description' => t('Used for "updates only" or "insert and update" methods. Not required if method type is "insert".
|
|
|
Feature data is stored in Chado with both a human-readable
|
|
|
name and a unique name. If the features in your FASTA file are uniquely identified using
|
|
|
a human-readable name then select the "Name" button. If your features are
|
|
|
@@ -267,7 +267,7 @@ function tripal_feature_fasta_load_form( ) {
|
|
|
}
|
|
|
|
|
|
/**
|
|
|
- *
|
|
|
+ * Validate the fasta loader job form
|
|
|
*
|
|
|
* @ingroup fasta_loader
|
|
|
*/
|
|
|
@@ -312,7 +312,7 @@ function tripal_feature_fasta_load_form_validate($form, &$form_state) {
|
|
|
if (!$re_name and $re_uname and strcmp($match_type, 'Name')==0) {
|
|
|
form_set_error('re_name', t("You must provide a regular expression to identify the sequence name"));
|
|
|
}
|
|
|
-
|
|
|
+
|
|
|
// check to see if the file is located local to Drupal
|
|
|
$fasta_file = trim($fasta_file);
|
|
|
$dfile = $_SERVER['DOCUMENT_ROOT'] . base_path() . $fasta_file;
|
|
|
@@ -367,7 +367,7 @@ function tripal_feature_fasta_load_form_validate($form, &$form_state) {
|
|
|
}
|
|
|
|
|
|
/**
|
|
|
- *
|
|
|
+ * Submit a fasta loading job
|
|
|
*
|
|
|
* @ingroup fasta_loader
|
|
|
*/
|
|
|
@@ -410,14 +410,49 @@ function tripal_feature_fasta_load_form_submit($form, &$form_state) {
|
|
|
$args = array($dfile, $organism_id, $type, $library_id, $re_name, $re_uname,
|
|
|
$re_accession, $db_id, $rel_type, $re_subject, $parent_type, $method,
|
|
|
$user->uid, $analysis_id, $match_type);
|
|
|
-
|
|
|
- $fname = preg_replace("/.*\/(.*)/", "$1", $dfile);
|
|
|
+
|
|
|
+ $fname = preg_replace("/.*\/(.*)/", "$1", $dfile);
|
|
|
tripal_add_job("Import FASTA file: $fname", 'tripal_feature',
|
|
|
'tripal_feature_load_fasta', $args, $user->uid);
|
|
|
}
|
|
|
|
|
|
/**
|
|
|
+ * Actually load a fasta file. This is the function called by tripal jobs
|
|
|
*
|
|
|
+ * @param $dfile
|
|
|
+ * The full path to the fasta file to load
|
|
|
+ * @param $organism_id
|
|
|
+ * The organism_id of the organism these features are from
|
|
|
+ * @param $type
|
|
|
+ * The type of features contained in the fasta file
|
|
|
+ * @param $library_id
|
|
|
+ * (Optional) The library_id to associate the features with
|
|
|
+ * @param $re_name
|
|
|
+ * A regular expression to extract the feature.name from the fasta header
|
|
|
+ * @param $re_uname
|
|
|
+ * A regular expression to extract the feature.uniquename from the fasta header
|
|
|
+ * @param $re_accession
|
|
|
+ * A regular expression to extract the accession of the feature.dbxref_id
|
|
|
+ * @param $db_id
|
|
|
+ * The db_id of the above dbxref
|
|
|
+ * @param $rel_type
|
|
|
+ * The type of relationship when creating a feature_relationship between this
|
|
|
+ * feature (object) and an extracted subject
|
|
|
+ * @param $re_subject
|
|
|
+ * The regular expression to extract the uniquename of the feature to be the subject
|
|
|
+ * of the above specified relationship
|
|
|
+ * @param $parent_type
|
|
|
+ * The type of the parent feature
|
|
|
+ * @param $method
|
|
|
+ * The method of feature adding. (ie: 'Insert only', 'Update only', 'Insert and update')
|
|
|
+ * @param $uid
|
|
|
+ * The user id of the user who submitted the job
|
|
|
+ * @param $analysis_id
|
|
|
+ * The analysis_id to associate the features in this fasta file with
|
|
|
+ * @param $match_type
|
|
|
+ * Whether to match existing features based on the 'Name' or 'Unique name'
|
|
|
+ * @param $job = NULL
|
|
|
+ * The tripal job
|
|
|
*
|
|
|
* @ingroup fasta_loader
|
|
|
*/
|
|
|
@@ -428,7 +463,7 @@ function tripal_feature_load_fasta($dfile, $organism_id, $type,
|
|
|
|
|
|
// begin the transaction
|
|
|
$connection = tripal_db_start_transaction();
|
|
|
-
|
|
|
+
|
|
|
// if we cannot get a connection then let the user know the loading will be slow
|
|
|
if (!$connection) {
|
|
|
print "A persistant connection was not obtained. Loading will be slow\n";
|
|
|
@@ -445,7 +480,7 @@ function tripal_feature_load_fasta($dfile, $organism_id, $type,
|
|
|
INNER JOIN {cv} CV on CVT.cv_id = CV.cv_id
|
|
|
LEFT JOIN {cvtermsynonym} CVTS on CVTS.cvterm_id = CVT.cvterm_id
|
|
|
WHERE cv.name = :cvname and (CVT.name = :name or CVTS.synonym = :synonym)";
|
|
|
- $cvterm = chado_query($cvtermsql, array(':cvname' => 'sequence', ':name' => $type, ':synonym' => $type))->fetchObject();
|
|
|
+ $cvterm = chado_query($cvtermsql, array(':cvname' => 'sequence', ':name' => $type, ':synonym' => $type))->fetchObject();
|
|
|
if (!$cvterm) {
|
|
|
tripal_report_error("T_fasta_loader", TRIPAL_ERROR, "Cannot find the term type: '%type'", array('%type' => $type));
|
|
|
return 0;
|
|
|
@@ -464,7 +499,7 @@ function tripal_feature_load_fasta($dfile, $organism_id, $type,
|
|
|
return 0;
|
|
|
}
|
|
|
}
|
|
|
-
|
|
|
+
|
|
|
print "Opening FASTA file $dfile\n";
|
|
|
|
|
|
//$lines = file($dfile, FILE_SKIP_EMPTY_LINES);
|
|
|
@@ -478,22 +513,22 @@ function tripal_feature_load_fasta($dfile, $organism_id, $type,
|
|
|
|
|
|
$name = '';
|
|
|
$uname = '';
|
|
|
- $residues = '';
|
|
|
+ $residues = '';
|
|
|
$interval = intval($filesize * 0.01);
|
|
|
if ($interval < 1) {
|
|
|
$interval = 1;
|
|
|
}
|
|
|
$inv_read = 0;
|
|
|
-
|
|
|
- // we need to get the table schema to make sure we don't overrun the
|
|
|
+
|
|
|
+ // we need to get the table schema to make sure we don't overrun the
|
|
|
// size of fields with what our regular expressions retrieve
|
|
|
$feature_tbl = chado_get_schema('feature');
|
|
|
$dbxref_tbl = chado_get_schema('dbxref');
|
|
|
|
|
|
- //foreach ($lines as $line_num => $line) {
|
|
|
+ //foreach ($lines as $line_num => $line) {
|
|
|
while ($line = fgets($fh)) {
|
|
|
$i++; // update the line count
|
|
|
- $num_read += drupal_strlen($line);
|
|
|
+ $num_read += drupal_strlen($line);
|
|
|
$intv_read += drupal_strlen($line);
|
|
|
|
|
|
// if we encounter a definition line then get the name, uniquename,
|
|
|
@@ -501,7 +536,7 @@ function tripal_feature_load_fasta($dfile, $organism_id, $type,
|
|
|
if (preg_match('/^>/', $line)) {
|
|
|
// if we have a feature name then we are starting a new sequence
|
|
|
// so lets handle the previous one before moving on
|
|
|
- if ($name or $uname) {
|
|
|
+ if ($name or $uname) {
|
|
|
tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id,
|
|
|
$accession, $subject, $rel_type, $parent_type, $analysis_id, $organism_id, $cvterm,
|
|
|
$source, $residues, $method, $re_name, $match_type, $parentcvterm, $relcvterm);
|
|
|
@@ -511,18 +546,18 @@ function tripal_feature_load_fasta($dfile, $organism_id, $type,
|
|
|
}
|
|
|
|
|
|
$line = preg_replace("/^>/", '', $line); // remove the > symbol from the defline
|
|
|
-
|
|
|
+
|
|
|
// get the feature name
|
|
|
if ($re_name) {
|
|
|
if (!preg_match("/$re_name/", $line, $matches)) {
|
|
|
tripal_report_error('trp-fasta', "ERROR: Regular expression for the feature name finds nothing. Line %line.", array('%line' => $i), 'error');
|
|
|
}
|
|
|
elseif (strlen($matches[1]) > $feature_tbl['fields']['name']['length']) {
|
|
|
- tripal_report_error('trp-fasta', "WARNING: Regular expression retrieves a value too long for the feature name. Line %line.", array('%line' => $i), 'error');
|
|
|
+ tripal_report_error('trp-fasta', "WARNING: Regular expression retrieves a value too long for the feature name. Line %line.", array('%line' => $i), 'error');
|
|
|
}
|
|
|
else {
|
|
|
$name = trim($matches[1]);
|
|
|
- }
|
|
|
+ }
|
|
|
}
|
|
|
else {
|
|
|
// if the match_type is name and no regular expression was provided
|
|
|
@@ -530,18 +565,18 @@ function tripal_feature_load_fasta($dfile, $organism_id, $type,
|
|
|
if (strcmp($match_type, 'Name')==0) {
|
|
|
if (preg_match("/^\s*(.*?)[\s\|].*$/", $line, $matches)) {
|
|
|
if (strlen($matches[1]) > $feature_tbl['fields']['name']['length']) {
|
|
|
- tripal_report_error('trp-fasta', "WARNING: Regular expression retrieves a feature name too long for the feature name. Line %line.", array('%line' => $i), 'error');
|
|
|
+ tripal_report_error('trp-fasta', "WARNING: Regular expression retrieves a feature name too long for the feature name. Line %line.", array('%line' => $i), 'error');
|
|
|
}
|
|
|
else {
|
|
|
$name = trim($matches[1]);
|
|
|
}
|
|
|
}
|
|
|
else {
|
|
|
- tripal_report_error('trp-fasta', "ERROR: Cannot find a feature name. Line %line.", array('%line' => $i), 'error');
|
|
|
+ tripal_report_error('trp-fasta', "ERROR: Cannot find a feature name. Line %line.", array('%line' => $i), 'error');
|
|
|
}
|
|
|
}
|
|
|
}
|
|
|
-
|
|
|
+
|
|
|
// get the feature unique name
|
|
|
if ($re_uname) {
|
|
|
if (!preg_match("/$re_uname/", $line, $matches)) {
|
|
|
@@ -557,14 +592,14 @@ function tripal_feature_load_fasta($dfile, $organism_id, $type,
|
|
|
$uname = trim($matches[1]);
|
|
|
}
|
|
|
else {
|
|
|
- tripal_report_error('trp-fasta', "ERROR: Cannot find a feature unique name. Line %line.", array('%line' => $i), 'error');
|
|
|
+ tripal_report_error('trp-fasta', "ERROR: Cannot find a feature unique name. Line %line.", array('%line' => $i), 'error');
|
|
|
}
|
|
|
}
|
|
|
}
|
|
|
// get the accession
|
|
|
preg_match("/$re_accession/", $line, $matches);
|
|
|
if (strlen($matches[1]) > $dbxref_tbl['fields']['accession']['length']) {
|
|
|
- tripal_report_error('trp-fasta', "WARNING: Regular expression retrieves an accession too long for the feature name. Cannot add cross reference. Line %line.", array('%line' => $i), 'warning');
|
|
|
+ tripal_report_error('trp-fasta', "WARNING: Regular expression retrieves an accession too long for the feature name. Cannot add cross reference. Line %line.", array('%line' => $i), 'warning');
|
|
|
}
|
|
|
else {
|
|
|
$accession = trim($matches[1]);
|
|
|
@@ -576,7 +611,7 @@ function tripal_feature_load_fasta($dfile, $organism_id, $type,
|
|
|
}
|
|
|
else {
|
|
|
$residues .= trim($line);
|
|
|
-
|
|
|
+
|
|
|
// update the job status every % features
|
|
|
if ($job and $intv_read >= $interval) {
|
|
|
$intv_read = 0;
|
|
|
@@ -585,25 +620,43 @@ function tripal_feature_load_fasta($dfile, $organism_id, $type,
|
|
|
print "Parsing Line $i (" . $percent . "%). Memory: " . number_format(memory_get_usage()) . " bytes. Current feature: $name\r";
|
|
|
}
|
|
|
else {
|
|
|
- print "Parsing Line $i (" . $percent . "%). Memory: " . number_format(memory_get_usage()) . " bytes. Current feature: $uname\r";
|
|
|
+ print "Parsing Line $i (" . $percent . "%). Memory: " . number_format(memory_get_usage()) . " bytes. Current feature: $uname\r";
|
|
|
}
|
|
|
tripal_set_job_progress($job, intval(($num_read / $filesize) * 100));
|
|
|
}
|
|
|
}
|
|
|
}
|
|
|
-
|
|
|
+
|
|
|
// now load the last sequence in the file
|
|
|
tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id,
|
|
|
$accession, $subject, $rel_type, $parent_type, $analysis_id, $organism_id, $cvterm,
|
|
|
$source, $residues, $method, $re_name, $match_type, $parentcvterm, $relcvterm);
|
|
|
-
|
|
|
+
|
|
|
// commit the transaction
|
|
|
tripal_db_commit_transaction();
|
|
|
print "\nDone\n";
|
|
|
}
|
|
|
|
|
|
/**
|
|
|
+ * A helper function for tripal_feature_load_fasta() to load a single feature
|
|
|
*
|
|
|
+ * @param $name
|
|
|
+ * @param $uname
|
|
|
+ * @param $db_id
|
|
|
+ * @param $accession
|
|
|
+ * @param $parent
|
|
|
+ * @param $rel_type
|
|
|
+ * @param $parent_type
|
|
|
+ * @param $analysis_id
|
|
|
+ * @param $organism_id
|
|
|
+ * @param $cvterm
|
|
|
+ * @param $source
|
|
|
+ * @param $residues
|
|
|
+ * @param $method
|
|
|
+ * @param $re_name
|
|
|
+ * @param $match_type
|
|
|
+ * @param $parentcvterm
|
|
|
+ * @param $relcvterm
|
|
|
*
|
|
|
* @ingroup fasta_loader
|
|
|
*/
|
|
|
@@ -616,44 +669,44 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
$values = array(
|
|
|
'organism_id' => $organism_id,
|
|
|
'name' => $name,
|
|
|
- 'type_id' => $cvterm->cvterm_id,
|
|
|
+ 'type_id' => $cvterm->cvterm_id,
|
|
|
);
|
|
|
$options = array('statement_name' => 'sel_feature_ornaty');
|
|
|
$results = chado_select_record('feature', array('feature_id'), $values, $options);
|
|
|
if (count($results) > 1) {
|
|
|
- tripal_report_error('T_fasta_loader', "Multiple features exist with the name '%name' of type
|
|
|
+ tripal_report_error('T_fasta_loader', "Multiple features exist with the name '%name' of type
|
|
|
'%type' for the organism. skipping", array('%name' => $name, '%type' => $type));
|
|
|
return 0;
|
|
|
- }
|
|
|
- if (count($results) == 1) {
|
|
|
+ }
|
|
|
+ if (count($results) == 1) {
|
|
|
$feature = $results[0];
|
|
|
- }
|
|
|
+ }
|
|
|
}
|
|
|
// check to see if this feature already exists if the match_type is 'Unique Name'
|
|
|
if (strcmp($match_type, 'Unique name')==0) {
|
|
|
$values = array(
|
|
|
'organism_id' => $organism_id,
|
|
|
'uniquename' => $uname,
|
|
|
- 'type_id' => $cvterm->cvterm_id,
|
|
|
+ 'type_id' => $cvterm->cvterm_id,
|
|
|
);
|
|
|
|
|
|
$options = array('statement_name' => 'sel_feature_oruqty');
|
|
|
$results = chado_select_record('feature', array('feature_id'), $values, $options);
|
|
|
if (count($results) > 1) {
|
|
|
- tripal_report_error('T_fasta_loader', "Multiple features exist with the name '%name' of type
|
|
|
+ tripal_report_error('T_fasta_loader', "Multiple features exist with the name '%name' of type
|
|
|
'%type' for the organism. skipping", array('%name' => $name, '%type' => $type));
|
|
|
return 0;
|
|
|
- }
|
|
|
- if (count($results) == 1) {
|
|
|
+ }
|
|
|
+ if (count($results) == 1) {
|
|
|
$feature = $results[0];
|
|
|
- }
|
|
|
-
|
|
|
- // if the feature exists but this is an "insert only" method then skip this feature
|
|
|
+ }
|
|
|
+
|
|
|
+ // if the feature exists but this is an "insert only" method then skip this feature
|
|
|
if ($feature and (strcmp($method, 'Insert only')==0)) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_WARNING, "Feature already exists '%name' ('%uname') while matching on %type. Skipping insert.",
|
|
|
+ tripal_report_error('T_fasta_loader', TRIPAL_WARNING, "Feature already exists '%name' ('%uname') while matching on %type. Skipping insert.",
|
|
|
array('%name' => $name, '%uname' => $uname, '%type' => drupal_strtolower($match_type)));
|
|
|
return 0;
|
|
|
- }
|
|
|
+ }
|
|
|
}
|
|
|
|
|
|
// if we don't have a feature and we're doing an insert then do the insert
|
|
|
@@ -666,7 +719,7 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
elseif (!$name) {
|
|
|
$name = $uname;
|
|
|
}
|
|
|
-
|
|
|
+
|
|
|
// insert the feature
|
|
|
$values = array(
|
|
|
'organism_id' => $organism_id,
|
|
|
@@ -680,44 +733,44 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
'is_obsolete' => 'FALSE',
|
|
|
);
|
|
|
$options = array('statement_name' => 'ins_feature_all');
|
|
|
- $success = chado_insert_record('feature', $values, $options);
|
|
|
+ $success = chado_insert_record('feature', $values, $options);
|
|
|
if (!$success) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to insert feature '%name (%uname)'",
|
|
|
+ tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to insert feature '%name (%uname)'",
|
|
|
array('%name' => $name, '%uname' => $numane));
|
|
|
return 0;
|
|
|
}
|
|
|
-
|
|
|
- // now get the feature we just inserted
|
|
|
+
|
|
|
+ // now get the feature we just inserted
|
|
|
$values = array(
|
|
|
'organism_id' => $organism_id,
|
|
|
'uniquename' => $uname,
|
|
|
- 'type_id' => $cvterm->cvterm_id,
|
|
|
+ 'type_id' => $cvterm->cvterm_id,
|
|
|
);
|
|
|
$options = array('statement_name' => 'sel_feature_oruqty');
|
|
|
$results = chado_select_record('feature', array('feature_id'), $values, $options);
|
|
|
if (count($results) == 1) {
|
|
|
$inserted = 1;
|
|
|
$feature = $results[0];
|
|
|
- }
|
|
|
+ }
|
|
|
else {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to retreive newly inserted feature '%name (%uname)'",
|
|
|
+ tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to retreive newly inserted feature '%name (%uname)'",
|
|
|
array('%name' => $name, '%uname' => $numane));
|
|
|
- return 0;
|
|
|
- }
|
|
|
+ return 0;
|
|
|
+ }
|
|
|
}
|
|
|
-
|
|
|
+
|
|
|
// if we don't have a feature and the user wants to do an update then fail
|
|
|
if (!$feature and (strcmp($method, 'Update only')==0 or drupal_strcmp($method, 'Insert and update')==0)) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to find feature '%name' ('%uname') while matching on " .
|
|
|
+ tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to find feature '%name' ('%uname') while matching on " .
|
|
|
drupal_strtolower($match_type), array('%name' => $name, '%uname' => $uname));
|
|
|
return 0;
|
|
|
}
|
|
|
-
|
|
|
+
|
|
|
// if we do have a feature and this is an update then proceed with the update
|
|
|
if ($feature and !$inserted and (strcmp($method, 'Update only')==0 or strcmp($method, 'Insert and update')==0)) {
|
|
|
// if the user wants to match on the Name field
|
|
|
if (strcmp($match_type, 'Name')==0) {
|
|
|
- // if we're matching on the name but do not have a unique name then we don't want to update the uniquename.
|
|
|
+ // if we're matching on the name but do not have a unique name then we don't want to update the uniquename.
|
|
|
$values = array();
|
|
|
if ($uname) {
|
|
|
// first check to make sure that by changing the unique name of this feature that we won't conflict with
|
|
|
@@ -725,17 +778,17 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
$values = array(
|
|
|
'organism_id' => $organism_id,
|
|
|
'uniquename' => $uname,
|
|
|
- 'type_id' => $cvterm->cvterm_id,
|
|
|
- );
|
|
|
+ 'type_id' => $cvterm->cvterm_id,
|
|
|
+ );
|
|
|
$options = array('statement_name' => 'sel_feature_oruqty');
|
|
|
$results = chado_select_record('feature', array('feature_id'), $values, $options);
|
|
|
if (count($results) > 0) {
|
|
|
- tripal_report_error('T_fasta_loader', "Cannot update the feature '%name' with a uniquename of '%uname' and type of '%type' as it
|
|
|
- conflicts with an existing feature with the same uniquename and type.",
|
|
|
+ tripal_report_error('T_fasta_loader', "Cannot update the feature '%name' with a uniquename of '%uname' and type of '%type' as it
|
|
|
+ conflicts with an existing feature with the same uniquename and type.",
|
|
|
array('%name' => $name, '%uname' => $uname, '%type' => $type));
|
|
|
return 0;
|
|
|
- }
|
|
|
-
|
|
|
+ }
|
|
|
+
|
|
|
// the changes to the uniquename don't conflict so proceed with the update
|
|
|
$values = array(
|
|
|
'uniquename' => $uname,
|
|
|
@@ -746,15 +799,15 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
'is_obsolete' => 'false',
|
|
|
);
|
|
|
$match = array(
|
|
|
- 'name' => $name,
|
|
|
+ 'name' => $name,
|
|
|
'organism_id' => $organism_id,
|
|
|
- 'type_id' => $cvterm->cvterm_id,
|
|
|
+ 'type_id' => $cvterm->cvterm_id,
|
|
|
);
|
|
|
- $options = array('statement_name' => 'upd_feature_resemdisis_naorty_un');
|
|
|
+ $options = array('statement_name' => 'upd_feature_resemdisis_naorty_un');
|
|
|
}
|
|
|
// if we do not have a new unique name then don't change the existing uniquename field
|
|
|
else {
|
|
|
- $values = array(
|
|
|
+ $values = array(
|
|
|
'residues' => $residues,
|
|
|
'seqlen' => drupal_strlen($residues),
|
|
|
'md5checksum' => md5($residues),
|
|
|
@@ -762,24 +815,24 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
'is_obsolete' => 'false',
|
|
|
);
|
|
|
$match = array(
|
|
|
- 'name' => $name,
|
|
|
+ 'name' => $name,
|
|
|
'organism_id' => $organism_id,
|
|
|
- 'type_id' => $cvterm->cvterm_id,
|
|
|
+ 'type_id' => $cvterm->cvterm_id,
|
|
|
);
|
|
|
- $options = array('statement_name' => 'upd_feature_unresemdisis_naorty');
|
|
|
+ $options = array('statement_name' => 'upd_feature_unresemdisis_naorty');
|
|
|
}
|
|
|
-
|
|
|
+
|
|
|
// perform the update
|
|
|
$success = chado_update_record('feature', $match, $values, $options);
|
|
|
if (!$success) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR,
|
|
|
- "Failed to update feature '%name' ('%name')",
|
|
|
+ tripal_report_error('T_fasta_loader', TRIPAL_ERROR,
|
|
|
+ "Failed to update feature '%name' ('%name')",
|
|
|
array('%name' => $name, '%uiname' => $uname));
|
|
|
return 0;
|
|
|
}
|
|
|
}
|
|
|
if (strcmp($match_type, 'Unique name')==0) {
|
|
|
- // if we're matching on the uniquename but do not have a new name then we don't want to update the name.
|
|
|
+ // if we're matching on the uniquename but do not have a new name then we don't want to update the name.
|
|
|
$values = array();
|
|
|
if ($name) {
|
|
|
$values = array(
|
|
|
@@ -792,14 +845,14 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
);
|
|
|
$match = array(
|
|
|
'uniquename' => $uname,
|
|
|
- 'organism_id' => $organism_id,
|
|
|
- 'type_id' => $cvterm->cvterm_id,
|
|
|
+ 'organism_id' => $organism_id,
|
|
|
+ 'type_id' => $cvterm->cvterm_id,
|
|
|
);
|
|
|
- $options = array('statement_name' => 'upd_feature_resemdisis_unorty_na');
|
|
|
+ $options = array('statement_name' => 'upd_feature_resemdisis_unorty_na');
|
|
|
}
|
|
|
// if we have a unique name then update it after matching by the name
|
|
|
else {
|
|
|
- $values = array(
|
|
|
+ $values = array(
|
|
|
'residues' => $residues,
|
|
|
'seqlen' => drupal_strlen($residues),
|
|
|
'md5checksum' => md5($residues),
|
|
|
@@ -807,21 +860,21 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
'is_obsolete' => 'false',
|
|
|
);
|
|
|
$match = array(
|
|
|
- 'uniquename' => $uname,
|
|
|
+ 'uniquename' => $uname,
|
|
|
'organism_id' => $organism_id,
|
|
|
- 'type_id' => $cvterm->cvterm_id,
|
|
|
+ 'type_id' => $cvterm->cvterm_id,
|
|
|
);
|
|
|
- $options = array('statement_name' => 'upd_feature_naresemdisis_unorty');
|
|
|
+ $options = array('statement_name' => 'upd_feature_naresemdisis_unorty');
|
|
|
}
|
|
|
$success = chado_update_record('feature', $match, $values, $options);
|
|
|
if (!$success) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to update feature '%name' ('%name')",
|
|
|
+ tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to update feature '%name' ('%name')",
|
|
|
array('%name' => $name, '%uiname' => $uname));
|
|
|
return 0;
|
|
|
}
|
|
|
- }
|
|
|
+ }
|
|
|
}
|
|
|
-
|
|
|
+
|
|
|
// add in the analysis link
|
|
|
if ($analysis_id) {
|
|
|
// if the association doens't alredy exist then add one
|
|
|
@@ -833,11 +886,11 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
$results = chado_select_record('analysisfeature', array('analysisfeature_id'), $values, $sel_options);
|
|
|
if (count($results) == 0) {
|
|
|
$ins_options = array('statement_name' => 'ins_analysisfeature_anfe');
|
|
|
- $success = chado_insert_record('analysisfeature', $values, $ins_options);
|
|
|
+ $success = chado_insert_record('analysisfeature', $values, $ins_options);
|
|
|
if (!$success) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to associate analysis and feature '%name' ('%name')",
|
|
|
+ tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to associate analysis and feature '%name' ('%name')",
|
|
|
array('%name' => $name, '%uname' => $uname));
|
|
|
- return 0;
|
|
|
+ return 0;
|
|
|
}
|
|
|
}
|
|
|
}
|
|
|
@@ -852,11 +905,11 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
$sel_options = array('statement_name' => 'sel_dbxref_dbac');
|
|
|
$results = chado_select_record('dbxref', array('dbxref_id'), $values, $sel_options);
|
|
|
// if the accession doesn't exist then add it
|
|
|
- if (count($results) == 0) {
|
|
|
+ if (count($results) == 0) {
|
|
|
$ins_options = array('statement_name' => 'ins_dbxref_dbac');
|
|
|
$results = chado_insert_record('dbxref', $values, $ins_options);
|
|
|
if (!$results) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to add database accession '%accession'",
|
|
|
+ tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to add database accession '%accession'",
|
|
|
array('%accession' => $accession));
|
|
|
return 0;
|
|
|
}
|
|
|
@@ -864,12 +917,12 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
if (count($results) == 1) {
|
|
|
$dbxref = $results[0];
|
|
|
}
|
|
|
- else {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to retreive newly inserted dbxref '%name (%uname)'",
|
|
|
+ else {
|
|
|
+ tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to retreive newly inserted dbxref '%name (%uname)'",
|
|
|
array('%name' => $name, '%uname' => $numane));
|
|
|
return 0;
|
|
|
}
|
|
|
- }
|
|
|
+ }
|
|
|
else {
|
|
|
$dbxref = $results[0];
|
|
|
}
|
|
|
@@ -881,15 +934,15 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
);
|
|
|
$sel_options = array('statement_name' => 'sel_featuredbxref_fedb');
|
|
|
$results = chado_select_record('feature_dbxref', array('feature_dbxref_id'), $values, $sel_options);
|
|
|
- if (count($results) == 0) {
|
|
|
+ if (count($results) == 0) {
|
|
|
$ins_options = array('statement_name' => 'ins_featuredbxref_fedb');
|
|
|
$success = chado_insert_record('feature_dbxref', $values, $ins_options);
|
|
|
if (!$success) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to add associate database accession '%accession' with feature",
|
|
|
+ tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to add associate database accession '%accession' with feature",
|
|
|
array('%accession' => $accession));
|
|
|
return 0;
|
|
|
}
|
|
|
- }
|
|
|
+ }
|
|
|
}
|
|
|
|
|
|
// now add in the relationship if one exists. If not, then add it
|
|
|
@@ -897,34 +950,33 @@ function tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id, $acce
|
|
|
$values = array(
|
|
|
'organism_id' => $organism_id,
|
|
|
'uniquename' => $parent,
|
|
|
- 'type_id' => $parentcvterm->cvterm_id,
|
|
|
+ 'type_id' => $parentcvterm->cvterm_id,
|
|
|
);
|
|
|
$options = array('statement_name' => 'sel_feature_oruqty');
|
|
|
$results = chado_select_record('feature', array('feature_id'), $values, $options);
|
|
|
if (count($results) != 1) {
|
|
|
- tripal_report_error('T_fasta_loader', "Cannot find a unique fature for the parent '%parent' of type
|
|
|
+ tripal_report_error('T_fasta_loader', "Cannot find a unique fature for the parent '%parent' of type
|
|
|
'%type' for the feature.", array('%parent' => $parent, '%type' => $parent_type));
|
|
|
return 0;
|
|
|
- }
|
|
|
+ }
|
|
|
$parent_feature = $results[0];
|
|
|
-
|
|
|
+
|
|
|
// check to see if the relationship already exists if not then add it
|
|
|
$values = array(
|
|
|
'subject_id' => $feature->feature_id,
|
|
|
'object_id' => $parent_feature->feature_id,
|
|
|
- 'type_id' => $relcvterm->cvterm_id,
|
|
|
+ 'type_id' => $relcvterm->cvterm_id,
|
|
|
);
|
|
|
$sel_options = array('statement_name' => 'sel_featurerelationship_suojty');
|
|
|
$results = chado_select_record('feature_relationship', array('feature_relationship_id'), $values, $sel_options);
|
|
|
- if (count($results) == 0) {
|
|
|
+ if (count($results) == 0) {
|
|
|
$ins_options = array('statement_name' => 'ins_featurerelationship_suojty');
|
|
|
$success = chado_insert_record('feature_relationship', $values, $ins_options);
|
|
|
if (!$success) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to add associate database accession '%accession' with feature",
|
|
|
+ tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to add associate database accession '%accession' with feature",
|
|
|
array('%accession' => $accession));
|
|
|
return 0;
|
|
|
}
|
|
|
- }
|
|
|
+ }
|
|
|
}
|
|
|
}
|
|
|
-
|
Lacey Sanderson