|
Tripal v1.0 (6.x-1.0)
|
|
Tripal v1.0 (6.x-1.0)
|
Functions | |
| tripal_feature_fasta_load_form () | |
| tripal_feature_fasta_load_form_validate ($form, &$form_state) | |
| tripal_feature_fasta_load_form_submit ($form, &$form_state) | |
| tripal_feature_load_fasta ($dfile, $organism_id, $type, $library_id, $re_name, $re_uname, $re_accession, $db_id, $rel_type, $re_subject, $parent_type, $method, $uid, $analysis_id, $match_type, $job=NULL) | |
| tripal_feature_fasta_loader_handle_feature ($name, $uname, $db_id, $accession, $parent, $rel_type, $parent_type, $analysis_id, $organism_id, $cvterm, $source, $residues, $method, $re_name, $match_type, $parentcvterm, $relcvterm) | |
Provides fasta loading functionality. Creates features based on their specification in a fasta file.
| tripal_feature_fasta_load_form | ( | ) |
Definition at line 23 of file fasta_loader.inc.
{
$form['fasta_file']= array(
'#type' => 'textfield',
'#title' => t('FASTA File'),
'#description' => t('Please enter the full system path for the FASTA file, or a path within the Drupal
installation (e.g. /sites/default/files/xyz.obo). The path must be accessible to the
server on which this Drupal instance is running.'),
'#required' => TRUE,
);
// get the list of organisms
$sql = "SELECT * FROM {organism} ORDER BY genus, species";
$org_rset = chado_query($sql);
$organisms = array();
$organisms[''] = '';
while ($organism = db_fetch_object($org_rset)) {
$organisms[$organism->organism_id] = "$organism->genus $organism->species ($organism->common_name)";
}
$form['organism_id'] = array(
'#title' => t('Organism'),
'#type' => t('select'),
'#description' => t("Choose the organism to which these sequences are associated"),
'#required' => TRUE,
'#options' => $organisms,
);
$form['seqtype']= array(
'#type' => 'textfield',
'#title' => t('Sequence Type'),
'#required' => TRUE,
'#description' => t('Please enter the Sequence Ontology (SO) term name that describes the sequences in the FASTA file (e.g. gene, mRNA, protein, etc...)'),
);
// get the list of organisms
$sql = "SELECT L.library_id, L.name, CVT.name as type
FROM {library} L
INNER JOIN {cvterm} CVT ON L.type_id = CVT.cvterm_id
ORDER BY name";
$lib_rset = chado_query($sql);
$libraries = array();
$libraries[''] = '';
while ($library = db_fetch_object($lib_rset)) {
$libraries[$library->library_id] = "$library->name ($library->type)";
}
// $form['library_id'] = array (
// '#title' => t('Library'),
// '#type' => t('select'),
// '#description' => t("Choose the library to which these sequences are associated "),
// '#required' => FALSE,
// '#options' => $libraries,
// '#weight' => 5,
// );
$form['method']= array(
'#type' => 'radios',
'#title' => 'Method',
'#required' => TRUE,
'#options' => array(
t('Insert only'),
t('Update only'),
t('Insert and update'),
),
'#description' => t('Select how features in the FASTA file are handled.
Select "Insert only" to insert the new features. If a feature already
exists with the same name or unique name and type then it is skipped.
Select "Update only" to only update featues that already exist in the
database. Select "Insert and Update" to insert features that do
not exist and upate those that do.'),
'#default_value' => 2,
);
$form['match_type']= array(
'#type' => 'radios',
'#title' => 'Name Match Type',
'#required' => TRUE,
'#options' => array(
t('Name'),
t('Unique name'),
),
'#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
uniquely identified using the unique name then select the "Unique name" button. If you
loaded your features first using the GFF loader then the unique name of each
features were indicated by the "ID=" attribute and the name by the "Name=" attribute.
By default, the FASTA loader will use the first word (character string
before the first space) as the name for your feature. If
this does not uniquely identify your feature consider specifying a regular expression in the advanced section below.
Additionally, you may import both a name and a unique name for each sequence using the advanced options.'),
'#default_value' => 1,
);
$form['analysis'] = array(
'#type' => 'fieldset',
'#title' => t('Analysis Used to Derive Features'),
'#collapsed' => TRUE
);
$form['analysis']['desc'] = array(
'#type' => 'markup',
'#value' => t("Why specify an analysis for a data load? All data comes
from some place, even if downloaded from Genbank. By specifying
analysis details for all data uploads, it allows an end user to reproduce the
data set, but at least indicates the source of the data."),
);
// get the list of organisms
$sql = "SELECT * FROM {analysis} ORDER BY name";
$org_rset = chado_query($sql);
$analyses = array();
$analyses[''] = '';
while ($analysis = db_fetch_object($org_rset)) {
$analyses[$analysis->analysis_id] = "$analysis->name ($analysis->program $analysis->programversion, $analysis->sourcename)";
}
$form['analysis']['analysis_id'] = array(
'#title' => t('Analysis'),
'#type' => t('select'),
'#description' => t("Choose the analysis to which these features are associated"),
'#required' => TRUE,
'#options' => $analyses,
);
// Advanced Options
$form['advanced'] = array(
'#type' => 'fieldset',
'#title' => t('Advanced Options'),
'#collapsible' => TRUE,
'#collapsed' => TRUE
);
$form['advanced']['re_help']= array(
'#type' => 'item',
'#value' => t('A regular expression is an advanced method for extracting information from a string of text.
Your FASTA file may contain both a human-readable name and a unique name for each sequence.
If you want to import
both the name and unique name for all sequences, then you must provide regular expressions
so that the loader knows how to separate them.
Otherwise the name and uniquename will be the same.
By default, this loader will use the first word in the definition
lines of the FASTA file
as the name or unique name of the feature.'),
);
$form['advanced']['re_name']= array(
'#type' => 'textfield',
'#title' => t('Regular expression for the name'),
'#required' => FALSE,
'#description' => t('Enter the regular expression that will extract the
feature name from the FASTA definition line. For example, for a
defintion line with a name and unique name separated by a bar \'|\' (>seqname|uniquename),
the regular expression for the name would be, "^(.*?)\|.*$".'),
);
$form['advanced']['re_uname']= array(
'#type' => 'textfield',
'#title' => t('Regular expression for the unique name'),
'#required' => FALSE,
'#description' => t('Enter the regular expression that will extract the
feature name from the FASTA definition line. For example, for a
defintion line with a name and unique name separated by a bar \'|\' (>seqname|uniquename),
the regular expression for the unique name would be "^.*?\|(.*)$").'),
);
// Advanced database cross-reference optoins
$form['advanced']['db'] = array(
'#type' => 'fieldset',
'#title' => t('External Database Reference'),
'#weight' => 6,
'#collapsed' => TRUE
);
$form['advanced']['db']['re_accession']= array(
'#type' => 'textfield',
'#title' => t('Regular expression for the accession'),
'#required' => FALSE,
'#description' => t('Enter the regular expression that will extract the accession for the external database for each feature from the FASTA definition line.'),
'#weight' => 2
);
// get the list of databases
$sql = "SELECT * FROM {db} ORDER BY name";
$db_rset = chado_query($sql);
$dbs = array();
$dbs[''] = '';
while ($db = db_fetch_object($db_rset)) {
$dbs[$db->db_id] = "$db->name";
}
$form['advanced']['db']['db_id'] = array(
'#title' => t('External Database'),
'#type' => t('select'),
'#description' => t("Plese choose an external database for which these sequences have a cross reference."),
'#required' => FALSE,
'#options' => $dbs,
'#weight' => 1,
);
$form['advanced']['relationship'] = array(
'#type' => 'fieldset',
'#title' => t('Relationships'),
'#weight' => 6,
'#collapsed' => TRUE
);
$rels = array();
$rels[''] = '';
$rels['part_of'] = 'part of';
$rels['derives_from'] = 'produced by';
// Advanced references options
$form['advanced']['relationship']['rel_type']= array(
'#title' => t('Relationship Type'),
'#type' => t('select'),
'#description' => t("Use this option to create associations, or relationships between the
features of this FASTA file and existing features in the database. For
example, to associate a FASTA file of peptides to existing genes or transcript sequence,
select the type 'produced by'. For a CDS sequences select the type 'part of'"),
'#required' => FALSE,
'#options' => $rels,
'#weight' => 5,
);
$form['advanced']['relationship']['re_subject']= array(
'#type' => 'textfield',
'#title' => t('Regular expression for the parent'),
'#required' => FALSE,
'#description' => t('Enter the regular expression that will extract the unique
name needed to identify the existing sequence for which the
relationship type selected above will apply.'),
'#weight' => 6
);
$form['advanced']['relationship']['parent_type']= array(
'#type' => 'textfield',
'#title' => t('Parent Type'),
'#required' => FALSE,
'#description' => t('Please enter the Sequence Ontology term for the parent. For example
if the FASTA file being loaded is a set of proteins that are
products of genes, then use the SO term \'gene\' or \'transcript\' or equivalent. However,
this type must match the type for already loaded features.'),
'#weight' => 7
);
$form['button'] = array(
'#type' => 'submit',
'#value' => t('Import FASTA file'),
'#weight' => 10,
);
return $form;
}
| tripal_feature_fasta_load_form_submit | ( | $ | form, |
| &$ | form_state | ||
| ) |
Definition at line 374 of file fasta_loader.inc.
{
global $user;
$dfile = $form_state['storage']['dfile'];
$organism_id = $form_state['values']['organism_id'];
$type = trim($form_state['values']['seqtype']);
$method = trim($form_state['values']['method']);
$match_type = trim($form_state['values']['match_type']);
$library_id = $form_state['values']['library_id'];
$re_name = trim($form_state['values']['re_name']);
$re_uname = trim($form_state['values']['re_uname']);
$re_accession = trim($form_state['values']['re_accession']);
$db_id = $form_state['values']['db_id'];
$rel_type = $form_state['values']['rel_type'];
$re_subject = trim($form_state['values']['re_subject']);
$parent_type = trim($form_state['values']['parent_type']);
$analysis_id = $form_state['values']['analysis_id'];
if ($method == 0) {
$method = 'Insert only';
}
if ($method == 1) {
$method = 'Update only';
}
if ($method == 2) {
$method = 'Insert and update';
}
if ($match_type == 0) {
$match_type = 'Name';
}
if ($match_type == 1) {
$match_type = 'Unique name';
}
$args = array($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);
tripal_add_job("Import FASTA file: $fname", 'tripal_feature',
'tripal_feature_load_fasta', $args, $user->uid);
}
| tripal_feature_fasta_load_form_validate | ( | $ | form, |
| &$ | form_state | ||
| ) |
Definition at line 274 of file fasta_loader.inc.
{
$fasta_file = trim($form_state['values']['fasta_file']);
$organism_id = $form_state['values']['organism_id'];
$type = trim($form_state['values']['seqtype']);
$method = trim($form_state['values']['method']);
$match_type = trim($form_state['values']['match_type']);
$library_id = $form_state['values']['library_id'];
$re_name = trim($form_state['values']['re_name']);
$re_uname = trim($form_state['values']['re_uname']);
$re_accession = trim($form_state['values']['re_accession']);
$db_id = $form_state['values']['db_id'];
$rel_type = $form_state['values']['rel_type'];
$re_subject = trim($form_state['values']['re_subject']);
$parent_type = trim($form_state['values']['parent_type']);
if ($method == 0) {
$method = 'Insert only';
}
if ($method == 1) {
$method = 'Update only';
}
if ($method == 2) {
$method = 'Insert and update';
}
if ($match_type == 0) {
$match_type = 'Name';
}
if ($match_type == 1) {
$match_type = 'Unique name';
}
if ($re_name and !$re_uname and strcmp($match_type, 'Unique name')==0) {
form_set_error('re_uname', t("You must provide a regular expression to identify the sequence unique name"));
}
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() . $fasta_file;
if (!file_exists($dfile)) {
// if not local to Drupal, the file must be someplace else, just use
// the full path provided
$dfile = $fasta_file;
}
if (!file_exists($dfile)) {
form_set_error('fasta_file', t("Cannot find the file on the system. Check that the file exists or that the web server has permissions to read the file."));
}
// make sure if a relationship is specified that all fields are provided.
if (($rel_type or $parent_type) and !$re_subject) {
form_set_error('re_subject', t("Please provide a regular expression for the parent"));
}
if (($rel_type or $re_subject) and !$parent_type) {
form_set_error('parent_type', t("Please provide a SO term for the parent"));
}
if (($parent_type or $re_subject) and !$rel_type) {
form_set_error('rel_type', t("Please select a relationship type"));
}
// make sure if a database is specified that all fields are provided
if ($db_id and !$re_accession) {
form_set_error('re_accession', t("Please provide a regular expression for the accession"));
}
if ($re_accession and !$db_id) {
form_set_error('db_id', t("Please select a database"));
}
// check to make sure the types exists
$cvtermsql = "SELECT CVT.cvterm_id
FROM {cvterm} CVT
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 = '%s' and (CVT.name = '%s' or CVTS.synonym = '%s')";
$cvterm = db_fetch_object(chado_query($cvtermsql, 'sequence', $type, $type));
if (!$cvterm) {
form_set_error('type', t("The Sequence Ontology (SO) term selected for the sequence type is not available in the database. Please check spelling or select another."));
}
if ($rel_type) {
$cvterm = db_fetch_object(chado_query($cvtermsql, 'sequence', $parent_type, $parent_type));
if (!$cvterm) {
form_set_error('parent_type', t("The Sequence Ontology (SO) term selected for the parent relationship is not available in the database. Please check spelling or select another."));
}
}
// check to make sure the 'relationship' and 'sequence' ontologies are loaded
$form_state['storage']['dfile'] = $dfile;
}
| tripal_feature_fasta_loader_handle_feature | ( | $ | name, |
| $ | uname, | ||
| $ | db_id, | ||
| $ | accession, | ||
| $ | parent, | ||
| $ | rel_type, | ||
| $ | parent_type, | ||
| $ | analysis_id, | ||
| $ | organism_id, | ||
| $ | cvterm, | ||
| $ | source, | ||
| $ | residues, | ||
| $ | method, | ||
| $ | re_name, | ||
| $ | match_type, | ||
| $ | parentcvterm, | ||
| $ | relcvterm | ||
| ) |
Definition at line 610 of file fasta_loader.inc.
{
// check to see if this feature already exists if the match_type is 'Name'
if (strcmp($match_type, 'Name')==0) {
$values = array(
'organism_id' => $organism_id,
'name' => $name,
'type_id' => $cvterm->cvterm_id,
);
$options = array('statement_name' => 'sel_feature_ornaty');
$results = tripal_core_chado_select('feature', array('feature_id'), $values, $options);
if (count($results) > 1) {
watchdog('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) {
$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,
);
$options = array('statement_name' => 'sel_feature_oruqty');
$results = tripal_core_chado_select('feature', array('feature_id'), $values, $options);
if (count($results) > 1) {
watchdog('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) {
$feature = $results[0];
}
// if the feature exists but this is an "insert only" method then skip this feature
if ($feature and (strcmp($method, 'Insert only')==0)) {
watchdog('T_fasta_loader', "Feature already exists '%name' ('%uname') while matching on %type. Skipping insert.",
array('%name' => $name, '%uname' => $uname, '%type' => drupal_strtolower($match_type)), WATCHDOG_WARNING);
return 0;
}
}
// if we don't have a feature and we're doing an insert then do the insert
$inserted = 0;
if (!$feature and (strcmp($method, 'Insert only')==0 or strcmp($method, 'Insert and update')==0)) {
// if we have a unique name but not a name then set them to be the same and vice versa
if (!$uname) {
$uname = $name;
}
elseif (!$name) {
$name = $uname;
}
// insert the feature
$values = array(
'organism_id' => $organism_id,
'name' => $name,
'uniquename' => $uname,
'residues' => $residues,
'seqlen' => drupal_strlen($residues),
'md5checksum' => md5($residues),
'type_id' => $cvterm->cvterm_id,
'is_analysis' => 'FALSE',
'is_obsolete' => 'FALSE',
);
$options = array('statement_name' => 'ins_feature_all');
$success = tripal_core_chado_insert('feature', $values, $options);
if (!$success) {
watchdog('T_fasta_loader', "Failed to insert feature '%name (%uname)'",
array('%name' => $name, '%uname' => $numane), WATCHDOG_ERROR);
return 0;
}
// now get the feature we just inserted
$values = array(
'organism_id' => $organism_id,
'uniquename' => $uname,
'type_id' => $cvterm->cvterm_id,
);
$options = array('statement_name' => 'sel_feature_oruqty');
$results = tripal_core_chado_select('feature', array('feature_id'), $values, $options);
if (count($results) == 1) {
$inserted = 1;
$feature = $results[0];
}
else {
watchdog('T_fasta_loader', "Failed to retreive newly inserted feature '%name (%uname)'",
array('%name' => $name, '%uname' => $numane), WATCHDOG_ERROR);
return 0;
}
}
// if we don't have a feature and the user wants to do an update then fail
if (!$feature and (strcmp($method, 'Update only')==0 or drupal_strcmp($method, 'Insert and update')==0)) {
watchdog('T_fasta_loader', "Failed to find feature '%name' ('%uname') while matching on " .
drupal_strtolower($match_type), array('%name' => $name, '%uname' => $uname), WATCHDOG_ERROR);
return 0;
}
// if we do have a feature and this is an update then proceed with the update
if ($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.
$values = array();
if ($uname) {
// first check to make sure that by changing the unique name of this feature that we won't conflict with
// another existing feature of the same name
$values = array(
'organism_id' => $organism_id,
'uniquename' => $uname,
'type_id' => $cvterm->cvterm_id,
);
$options = array('statement_name' => 'sel_feature_oruqty');
$results = tripal_core_chado_select('feature', array('feature_id'), $values, $options);
if (count($results) > 0) {
watchdog('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,
'residues' => $residues,
'seqlen' => drupal_strlen($residues),
'md5checksum' => md5($residues),
'is_analysis' => 'false',
'is_obsolete' => 'false',
);
$match = array(
'name' => $name,
'organism_id' => $organism_id,
'type_id' => $cvterm->cvterm_id,
);
$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(
'residues' => $residues,
'seqlen' => drupal_strlen($residues),
'md5checksum' => md5($residues),
'is_analysis' => 'false',
'is_obsolete' => 'false',
);
$match = array(
'name' => $name,
'organism_id' => $organism_id,
'type_id' => $cvterm->cvterm_id,
);
$options = array('statement_name' => 'upd_feature_unresemdisis_naorty');
}
// perform the update
$success = tripal_core_chado_update('feature', $match, $values, $options);
if (!$success) {
watchdog('T_fasta_loader', "Failed to update feature '%name' ('%name')",
array('%name' => $name, '%uiname' => $uname), WATCHDOG_ERROR);
return 0;
}
}
if (strcmp($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.
$values = array();
if ($name) {
$values = array(
'name' => $name,
'residues' => $residues,
'seqlen' => drupal_strlen($residues),
'md5checksum' => md5($residues),
'is_analysis' => 'false',
'is_obsolete' => 'false',
);
$match = array(
'uniquename' => $uname,
'organism_id' => $organism_id,
'type_id' => $cvterm->cvterm_id,
);
$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(
'residues' => $residues,
'seqlen' => drupal_strlen($residues),
'md5checksum' => md5($residues),
'is_analysis' => 'false',
'is_obsolete' => 'false',
);
$match = array(
'uniquename' => $uname,
'organism_id' => $organism_id,
'type_id' => $cvterm->cvterm_id,
);
$options = array('statement_name' => 'upd_feature_naresemdisis_unorty');
}
$success = tripal_core_chado_update('feature', $match, $values, $options);
if (!$success) {
watchdog('T_fasta_loader', "Failed to update feature '%name' ('%name')",
array('%name' => $name, '%uiname' => $uname), WATCHDOG_ERROR);
return 0;
}
}
}
// add in the analysis link
if ($analysis_id) {
// if the association doens't alredy exist then add one
$values = array(
'analysis_id' => $analysis_id,
'feature_id' => $feature->feature_id,
);
$sel_options = array('statement_name' => 'sel_analysisfeature_anfe');
$results = tripal_core_chado_select('analysisfeature', array('analysisfeature_id'), $values, $sel_options);
if (count($results) == 0) {
$ins_options = array('statement_name' => 'ins_analysisfeature_anfe');
$success = tripal_core_chado_insert('analysisfeature', $values, $ins_options);
if (!$success) {
watchdog('T_fasta_loader', "Failed to associate analysis and feature '%name' ('%name')",
array('%name' => $name, '%uname' => $uname), WATCHDOG_ERROR);
return 0;
}
}
}
// now add the database cross reference
if ($db_id) {
// check to see if this accession reference exists, if not add it
$values = array(
'db_id' => $db_id,
'accession' => $accession
);
$sel_options = array('statement_name' => 'sel_dbxref_dbac');
$results = tripal_core_chado_select('dbxref', array('dbxref_id'), $values, $sel_options);
// if the accession doesn't exist then add it
if (count($results) == 0) {
$ins_options = array('statement_name' => 'ins_dbxref_dbac');
$results = tripal_core_chado_insert('dbxref', $values, $ins_options);
if (!$results) {
watchdog('T_fasta_loader', "Failed to add database accession '%accession'",
array('%accession' => $accession), WATCHDOG_ERROR);
return 0;
}
$results = tripal_core_chado_select('dbxref', array('dbxref_id'), $values, $sel_options);
if (count($results) == 1) {
$dbxref = $results[0];
}
else {
watchdog('T_fasta_loader', "Failed to retreive newly inserted dbxref '%name (%uname)'",
array('%name' => $name, '%uname' => $numane), WATCHDOG_ERROR);
return 0;
}
}
else {
$dbxref = $results[0];
}
// check to see if the feature dbxref record exists if not, then add it
$values = array(
'feature_id' => $feature->feature_id,
'dbxref_id' => $dbxref->dbxref_id
);
$sel_options = array('statement_name' => 'sel_featuredbxref_fedb');
$results = tripal_core_chado_select('feature_dbxref', array('feature_dbxref_id'), $values, $sel_options);
if (count($results) == 0) {
$ins_options = array('statement_name' => 'ins_featuredbxref_fedb');
$success = tripal_core_chado_insert('feature_dbxref', $values, $ins_options);
if (!$success) {
watchdog('T_fasta_loader', "Failed to add associate database accession '%accession' with feature",
array('%accession' => $accession), WATCHDOG_ERROR);
return 0;
}
}
}
// now add in the relationship if one exists. If not, then add it
if ($rel_type) {
$values = array(
'organism_id' => $organism_id,
'uniquename' => $parent,
'type_id' => $parentcvterm->cvterm_id,
);
$options = array('statement_name' => 'sel_feature_oruqty');
$results = tripal_core_chado_select('feature', array('feature_id'), $values, $options);
if (count($results) != 1) {
watchdog('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,
);
$sel_options = array('statement_name' => 'sel_featurerelationship_suojty');
$results = tripal_core_chado_select('feature_relationship', array('feature_relationship_id'), $values, $sel_options);
if (count($results) == 0) {
$ins_options = array('statement_name' => 'ins_featurerelationship_suojty');
$success = tripal_core_chado_insert('feature_relationship', $values, $ins_options);
if (!$success) {
watchdog('T_fasta_loader', "Failed to add associate database accession '%accession' with feature",
array('%accession' => $accession), WATCHDOG_ERROR);
return 0;
}
}
}
}
| tripal_feature_load_fasta | ( | $ | dfile, |
| $ | organism_id, | ||
| $ | type, | ||
| $ | library_id, | ||
| $ | re_name, | ||
| $ | re_uname, | ||
| $ | re_accession, | ||
| $ | db_id, | ||
| $ | rel_type, | ||
| $ | re_subject, | ||
| $ | parent_type, | ||
| $ | method, | ||
| $ | uid, | ||
| $ | analysis_id, | ||
| $ | match_type, | ||
| $ | job = NULL |
||
| ) |
Definition at line 424 of file fasta_loader.inc.
{
// 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";
}
else {
print "\nNOTE: Loading of this FASTA file is performed using a database transaction. \n" .
"If the load fails or is terminated prematurely then the entire set of \n" .
"insertions/updates is rolled back and will not be found in the database\n\n";
}
// first get the type for this sequence
$cvtermsql = "SELECT CVT.cvterm_id
FROM {cvterm} CVT
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 = '%s' and (CVT.name = '%s' or CVTS.synonym = '%s')";
$cvterm = db_fetch_object(chado_query($cvtermsql, 'sequence', $type, $type));
if (!$cvterm) {
watchdog("T_fasta_loader", "Cannot find the term type: '%type'", array('%type' => $type), WATCHDOG_ERROR);
return 0;
}
if ($parent_type) {
$parentcvterm = db_fetch_object(chado_query($cvtermsql, 'sequence', $parent_type, $parent_type));
if (!$parentcvterm) {
watchdog("T_fasta_loader", "Cannot find the paretne term type: '%type'", array('%type' => $parentcvterm), WATCHDOG_ERROR);
return 0;
}
}
if ($rel_type) {
$relcvterm = db_fetch_object(chado_query($cvtermsql, 'relationship', $rel_type, $rel_type));
if (!$relcvterm) {
watchdog("T_fasta_loader", "Cannot find the relationship term type: '%type'", array('%type' => $relcvterm), WATCHDOG_ERROR);
return 0;
}
}
print "Opening FASTA file $dfile\n";
//$lines = file($dfile, FILE_SKIP_EMPTY_LINES);
$fh = fopen($dfile, 'r');
if (!$fh) {
watchdog('T_fasta_loader', "cannot open file: %dfile", array('%dfile' => $dfile), WATCHDOG_ERROR);
return 0;
}
$filesize = filesize($dfile);
$i = 0;
$name = '';
$uname = '';
$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
// size of fields with what our regular expressions retrieve
$feature_tbl = tripal_core_get_chado_table_schema('feature');
$dbxref_tbl = tripal_core_get_chado_table_schema('dbxref');
//foreach ($lines as $line_num => $line) {
while ($line = fgets($fh)) {
$i++; // update the line count
$num_read += drupal_strlen($line);
$intv_read += drupal_strlen($line);
// if we encounter a definition line then get the name, uniquename,
// accession and relationship subject from the definition line
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 $uname) {
tripal_feature_fasta_loader_handle_feature($name, $uname, $db_id,
$accession, $subject, $rel_type, $parent_type, $analysis_id, $organism_id, $cvterm,
$source, $residues, $method, $re_name, $match_type, $parentcvterm, $relcvterm);
$residues = '';
$name = '';
$uname = '';
}
$line = preg_replace("/^>/", '', $line); // remove the > symbol from the defline
// get the feature name
if ($re_name) {
if (!preg_match("/$re_name/", $line, $matches)) {
watchdog('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']) {
watchdog('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
// then use the first word as the name, otherwise we don't set the name
if (strcmp($match_type, 'Name')==0) {
if(preg_match("/^\s*(.*?)[\s\|].*$/", $line, $matches)){
if (strlen($matches[1]) > $feature_tbl['fields']['name']['length']) {
watchdog('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 {
watchdog('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)) {
watchdog('trp-fasta', "ERROR: Regular expression for the feature unique name finds nothing. Line %line.", array('%line' => $i), 'error');
}
$uname = trim($matches[1]);
}
else {
// if the match_type is name and no regular expression was provided
// then use the first word as the name, otherwise, we don't set the unqiuename
if (strcmp($match_type, 'Unique name')==0) {
if(preg_match("/^\s*(.*?)[\s\|].*$/", $line, $matches)){
$uname = trim($matches[1]);
}
else {
watchdog('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']) {
watchdog('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]);
}
// get the relationship subject
preg_match("/$re_subject/", $line, $matches);
$subject = trim($matches[1]);
}
else {
$residues .= trim($line);
// update the job status every % features
if ($job and $intv_read >= $interval) {
$intv_read = 0;
$percent = sprintf("%.2f", ($num_read / $filesize) * 100);
if ($name) {
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: $uname\r";
}
tripal_job_set_progress($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";
}
1.7.4