|
|
@@ -46,9 +46,12 @@ class FASTAImporter extends TripalImporter {
|
|
|
while ($organism = $org_rset->fetchObject()) {
|
|
|
$organisms[$organism->organism_id] = "$organism->genus $organism->species ($organism->common_name)";
|
|
|
}
|
|
|
- $form['organism_id'] = array('#title' => t('Organism'),'#type' => t('select'),
|
|
|
+ $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
|
|
|
+ '#required' => TRUE,
|
|
|
+ '#options' => $organisms
|
|
|
);
|
|
|
|
|
|
// get the sequence ontology CV ID
|
|
|
@@ -56,102 +59,97 @@ class FASTAImporter extends TripalImporter {
|
|
|
$cv = chado_select_record('cv', array('cv_id'), $values);
|
|
|
$cv_id = $cv[0]->cv_id;
|
|
|
|
|
|
- $form['seqtype'] = array('#type' => 'textfield','#title' => t('Sequence Type'),
|
|
|
+ $form['seqtype'] = array(
|
|
|
+ '#type' => 'textfield',
|
|
|
+ '#title' => t('Sequence Type'),
|
|
|
'#required' => TRUE,
|
|
|
'#description' => t('Please enter the Sequence Ontology (SO) term name that describes the sequences in the FASTA file (e.g. gene, mRNA, polypeptide, etc...)'),
|
|
|
'#autocomplete_path' => "admin/tripal/storage/chado/auto_name/cvterm/$cv_id"
|
|
|
);
|
|
|
|
|
|
- $form['method'] = array('#type' => 'radios','#title' => 'Method','#required' => TRUE,
|
|
|
- '#options' => array(t('Insert only'),t('Update only'),t('Insert and update')
|
|
|
+ $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
|
|
|
+ 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.'),
|
|
|
+ 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'),
|
|
|
+ // Additional Options
|
|
|
+ $form['additional'] = array(
|
|
|
+ '#type' => 'fieldset',
|
|
|
+ '#title' => t('Additional Options'),
|
|
|
+ '#collapsible' => TRUE,
|
|
|
'#collapsed' => TRUE
|
|
|
);
|
|
|
- $form['analysis']['desc'] = array(
|
|
|
- '#markup' => 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 = $org_rset->fetchObject()) {
|
|
|
- $analyses[$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',
|
|
|
+ $form['additional']['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.')
|
|
|
+ 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',
|
|
|
+ $form['additional']['re_name'] = array('#type' => 'textfield',
|
|
|
'#title' => t('Regular expression for the name'),'#required' => FALSE,
|
|
|
'#description' => t('Enter the regular expression that will extract the
|
|
|
- feature name from the FASTA definition line. For example, for a
|
|
|
- defintion line with a name and unique name separated by a bar \'|\' (>seqname|uniquename),
|
|
|
- the regular expression for the name would be, "^(.*?)\|.*$". All FASTA
|
|
|
- definition lines begin with the ">" symbol. You do not need to incldue
|
|
|
- this symbol in your regular expression.')
|
|
|
+ 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, "^(.*?)\|.*$". All FASTA
|
|
|
+ definition lines begin with the ">" symbol. You do not need to incldue
|
|
|
+ this symbol in your regular expression.')
|
|
|
);
|
|
|
- $form['advanced']['re_uname'] = array('#type' => 'textfield',
|
|
|
+ $form['additional']['re_uname'] = array('#type' => 'textfield',
|
|
|
'#title' => t('Regular expression for the unique name'),'#required' => FALSE,
|
|
|
'#description' => t('Enter the regular expression that will extract the
|
|
|
- feature name from the FASTA definition line. For example, for a
|
|
|
- defintion line with a name and unique name separated by a bar \'|\' (>seqname|uniquename),
|
|
|
- the regular expression for the unique name would be "^.*?\|(.*)$"). All FASTA
|
|
|
- definition lines begin with the ">" symbol. You do not need to incldue
|
|
|
- this symbol in your regular expression.')
|
|
|
+ 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 "^.*?\|(.*)$"). All FASTA
|
|
|
+ definition lines begin with the ">" symbol. You do not need to incldue
|
|
|
+ this symbol in your regular expression.')
|
|
|
);
|
|
|
|
|
|
// Advanced database cross reference options.
|
|
|
- $form['advanced']['db'] = array('#type' => 'fieldset',
|
|
|
- '#title' => t('External Database Reference'),'#weight' => 6,'#collapsed' => TRUE
|
|
|
+ $form['additional']['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,
|
|
|
+ $form['additional']['db']['re_accession'] = array(
|
|
|
+ '#type' => 'textfield',
|
|
|
+ '#title' => t('Regular expression for the accession'),
|
|
|
+ '#required' => FALSE,
|
|
|
'#description' => t('Enter the regular expression that will extract the accession for the external database for each feature from the FASTA definition line.'),
|
|
|
'#weight' => 2
|
|
|
);
|
|
|
@@ -164,13 +162,18 @@ class FASTAImporter extends TripalImporter {
|
|
|
while ($db = $db_rset->fetchObject()) {
|
|
|
$dbs[$db->db_id] = "$db->name";
|
|
|
}
|
|
|
- $form['advanced']['db']['db_id'] = array('#title' => t('External Database'),
|
|
|
+ $form['additional']['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
|
|
|
+ '#required' => FALSE,
|
|
|
+ '#options' => $dbs,
|
|
|
+ '#weight' => 1
|
|
|
);
|
|
|
|
|
|
- $form['advanced']['relationship'] = array('#type' => 'fieldset','#title' => t('Relationships'),
|
|
|
+ $form['additional']['relationship'] = array(
|
|
|
+ '#type' => 'fieldset',
|
|
|
+ '#title' => t('Relationships'),
|
|
|
'#weight' => 6,'#collapsed' => TRUE
|
|
|
);
|
|
|
$rels = array();
|
|
|
@@ -179,22 +182,30 @@ class FASTAImporter extends TripalImporter {
|
|
|
$rels['derives_from'] = 'produced by (derives from)';
|
|
|
|
|
|
// Advanced references options
|
|
|
- $form['advanced']['relationship']['rel_type'] = array('#title' => t('Relationship Type'),
|
|
|
+ $form['additional']['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
|
|
|
+ '#required' => FALSE,
|
|
|
+ '#options' => $rels,
|
|
|
+ '#weight' => 5
|
|
|
);
|
|
|
- $form['advanced']['relationship']['re_subject'] = array('#type' => 'textfield',
|
|
|
- '#title' => t('Regular expression for the parent'),'#required' => FALSE,
|
|
|
+ $form['additional']['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
|
|
|
+ relationship type selected above will apply.'),
|
|
|
+ '#weight' => 6
|
|
|
);
|
|
|
- $form['advanced']['relationship']['parent_type'] = array('#type' => 'textfield',
|
|
|
- '#title' => t('Parent Type'),'#required' => FALSE,
|
|
|
+ $form['additional']['relationship']['parent_type'] = array(
|
|
|
+ '#type' => 'textfield',
|
|
|
+ '#title' => t('Parent Type'),
|
|
|
+ '#required' => FALSE,
|
|
|
'#description' => t('Please enter the Sequence Ontology term for the parent. For example
|
|
|
if the FASTA file being loaded is a set of proteins that are
|
|
|
products of genes, then use the SO term \'gene\' or \'transcript\' or equivalent. However,
|
|
|
@@ -305,7 +316,7 @@ class FASTAImporter extends TripalImporter {
|
|
|
/**
|
|
|
* @see TripalImporter::run()
|
|
|
*/
|
|
|
- public function run($details, $job_id = NULL) {
|
|
|
+ public function run($details) {
|
|
|
$arguments = $details->arguments;
|
|
|
|
|
|
$dfile = $arguments['file_path'];
|
|
|
@@ -324,7 +335,7 @@ class FASTAImporter extends TripalImporter {
|
|
|
|
|
|
$this->load($dfile, $organism_id, $type, $re_name, $re_uname, $re_accession,
|
|
|
$db_id, $rel_type, $re_subject, $parent_type, $method, $uid, $analysis_id,
|
|
|
- $match_type, $job_id);
|
|
|
+ $match_type);
|
|
|
}
|
|
|
/**
|
|
|
* Load a fasta file.
|
|
|
@@ -361,25 +372,21 @@ class FASTAImporter extends TripalImporter {
|
|
|
* 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 $job
|
|
|
- * The tripal job
|
|
|
*/
|
|
|
private function loadFasta($dfile, $organism_id, $type, $re_name, $re_uname, $re_accession,
|
|
|
- $db_id, $rel_type, $re_subject, $parent_type, $method, $uid, $analysis_id, $match_type,
|
|
|
- $job = NULL) {
|
|
|
+ $db_id, $rel_type, $re_subject, $parent_type, $method, $uid, $analysis_id, $match_type) {
|
|
|
|
|
|
// 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 = :cvname and (CVT.name = :name or CVTS.synonym = :synonym)
|
|
|
- ";
|
|
|
+ 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 = :cvname and (CVT.name = :name or CVTS.synonym = :synonym)
|
|
|
+ ";
|
|
|
$cvterm = chado_query($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));
|
|
|
+ $this->logMessage("Cannot find the term type: '%type'", array('%type' => $type), TRIPAL_ERROR);
|
|
|
return 0;
|
|
|
}
|
|
|
|
|
|
@@ -387,9 +394,9 @@ class FASTAImporter extends TripalImporter {
|
|
|
if ($parent_type) {
|
|
|
$parentcvterm = chado_query($cvtermsql, array(':cvname' => 'sequence', ':name' => $parent_type,':synonym' => $parent_type))->fetchObject();
|
|
|
if (!$parentcvterm) {
|
|
|
- tripal_report_error("T_fasta_loader", TRIPAL_ERROR, "Cannot find the paretne term type: '%type'", array(
|
|
|
+ $this->logMessage("Cannot find the paretne term type: '%type'", array(
|
|
|
'%type' => $parentcvterm
|
|
|
- ));
|
|
|
+ ), TRIPAL_ERROR);
|
|
|
return 0;
|
|
|
}
|
|
|
}
|
|
|
@@ -398,9 +405,9 @@ class FASTAImporter extends TripalImporter {
|
|
|
if ($rel_type) {
|
|
|
$relcvterm = chado_query($cvtermsql, array(':cvname' => 'sequence',':name' => $rel_type,':synonym' => $rel_type))->fetchObject();
|
|
|
if (!$relcvterm) {
|
|
|
- tripal_report_error("T_fasta_loader", TRIPAL_ERROR, "Cannot find the relationship term type: '%type'", array(
|
|
|
+ $this->logMessage("Cannot find the relationship term type: '%type'", array(
|
|
|
'%type' => $relcvterm
|
|
|
- ));
|
|
|
+ ), TRIPAL_ERROR);
|
|
|
return 0;
|
|
|
}
|
|
|
}
|
|
|
@@ -410,22 +417,16 @@ class FASTAImporter extends TripalImporter {
|
|
|
$feature_tbl = chado_get_schema('feature');
|
|
|
$dbxref_tbl = chado_get_schema('dbxref');
|
|
|
|
|
|
- print "Step 1: finding sequences\n";
|
|
|
+ $this->logMessage(t("Step 1: finding sequences\n"));
|
|
|
$filesize = filesize($dfile);
|
|
|
+ $this->setTotalItems($filesize);
|
|
|
$fh = fopen($dfile, 'r');
|
|
|
if (!$fh) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "cannot open file: %dfile", array(
|
|
|
+ $this->logMessage("Cannot open file: %dfile", array(
|
|
|
'%dfile' => $dfile
|
|
|
- ));
|
|
|
+ ), TRIPAL_ERROR);
|
|
|
return 0;
|
|
|
}
|
|
|
-
|
|
|
- // Calculate the interval at which we will print to the screen that status.
|
|
|
- $interval = intval($filesize * 0.01);
|
|
|
- if ($interval < 1) {
|
|
|
- $interval = 1;
|
|
|
- }
|
|
|
- $inv_read = 0;
|
|
|
$num_read = 0;
|
|
|
|
|
|
// Iterate through the lines of the file. Keep a record for
|
|
|
@@ -436,7 +437,6 @@ class FASTAImporter extends TripalImporter {
|
|
|
$set_start = FALSE;
|
|
|
while ($line = fgets($fh)) {
|
|
|
$num_read += strlen($line);
|
|
|
- $intv_read += strlen($line);
|
|
|
|
|
|
// If we encounter a definition line then get the name, uniquename,
|
|
|
// accession and relationship subject from the definition line.
|
|
|
@@ -448,14 +448,12 @@ class FASTAImporter extends TripalImporter {
|
|
|
// Get the feature name if a regular expression is provided.
|
|
|
if ($re_name) {
|
|
|
if (!preg_match("/$re_name/", $defline, $matches)) {
|
|
|
- tripal_report_error('trp-fasta', "ERROR: Regular expression for the feature name finds nothing. Line %line.", array(
|
|
|
- '%line' => $i
|
|
|
- ), 'error');
|
|
|
+ $this->logMessage("Regular expression for the feature name finds nothing. Line %line.",
|
|
|
+ array('%line' => $i), TRIPAL_ERROR);
|
|
|
}
|
|
|
elseif (strlen($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');
|
|
|
+ $this->logMessage("Regular expression retrieves a value too long for the feature name. Line %line.",
|
|
|
+ array('%line' => $i), TRIPAL_WARNING);
|
|
|
}
|
|
|
else {
|
|
|
$name = trim($matches[1]);
|
|
|
@@ -466,24 +464,23 @@ class FASTAImporter extends TripalImporter {
|
|
|
elseif (strcmp($match_type, 'Name') == 0) {
|
|
|
if (preg_match("/^\s*(.*?)[\s\|].*$/", $defline, $matches)) {
|
|
|
if (strlen($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');
|
|
|
+ $this->logMessage("Regular expression retrieves a feature name too long for the feature name. Line %line.",
|
|
|
+ array('%line' => $i), TRIPAL_WARNING);
|
|
|
}
|
|
|
else {
|
|
|
$name = trim($matches[1]);
|
|
|
}
|
|
|
}
|
|
|
else {
|
|
|
- tripal_report_error('trp-fasta', "ERROR: Cannot find a feature name. Line %line.", array(
|
|
|
- '%line' => $i), 'error');
|
|
|
+ $this->logMessage("Cannot find a feature name. Line %line.", array('%line' => $i), TRIPAL_WARNING);
|
|
|
}
|
|
|
}
|
|
|
|
|
|
// Get the feature uniquename if a regular expression is provided.
|
|
|
if ($re_uname) {
|
|
|
if (!preg_match("/$re_uname/", $defline, $matches)) {
|
|
|
- tripal_report_error('trp-fasta', "ERROR: Regular expression for the feature unique name finds nothing. Line %line.", array(
|
|
|
- '%line' => $i), 'error');
|
|
|
+ $this->logMessage("Regular expression for the feature unique name finds nothing. Line %line.",
|
|
|
+ array('%line' => $i), TRIPAL_ERROR);
|
|
|
}
|
|
|
$uname = trim($matches[1]);
|
|
|
}
|
|
|
@@ -495,17 +492,16 @@ class FASTAImporter extends TripalImporter {
|
|
|
$uname = trim($matches[1]);
|
|
|
}
|
|
|
else {
|
|
|
- tripal_report_error('trp-fasta', "ERROR: Cannot find a feature unique name. Line %line.", array(
|
|
|
- '%line' => $i), 'error');
|
|
|
+ $this->logMessage("Cannot find a feature unique name. Line %line.",
|
|
|
+ array('%line' => $i), TRIPAL_ERROR);
|
|
|
}
|
|
|
}
|
|
|
|
|
|
// Get the accession if a regular expression is provided.
|
|
|
preg_match("/$re_accession/", $defline, $matches);
|
|
|
if (strlen($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');
|
|
|
+ $this->logMessage("Regular expression retrieves an accession too long for the feature name. " .
|
|
|
+ "Cannot add cross reference. Line %line.", array('%line' => $i), TRIPAL_WARNING);
|
|
|
}
|
|
|
else {
|
|
|
$accession = trim($matches[1]);
|
|
|
@@ -535,36 +531,18 @@ class FASTAImporter extends TripalImporter {
|
|
|
// ending position
|
|
|
$prev_pos = ftell($fh);
|
|
|
|
|
|
- // update the job status every % bytes
|
|
|
- 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.\r";
|
|
|
- }
|
|
|
- else {
|
|
|
- print "Parsing Line $i (" . $percent . "%). Memory: " . number_format(memory_get_usage()) .
|
|
|
- " bytes.\r";
|
|
|
- }
|
|
|
- tripal_set_job_progress($job, intval(($num_read / $filesize) * 100));
|
|
|
- }
|
|
|
}
|
|
|
- $percent = sprintf("%.2f", ($num_read / $filesize) * 100);
|
|
|
- print "Parsing Line $i (" . $percent . "%). Memory: " . number_format(memory_get_usage()) .
|
|
|
- " bytes.\r";
|
|
|
- tripal_set_job_progress($job, 50);
|
|
|
|
|
|
// Set the end position for the last sequence.
|
|
|
$seqs[$num_seqs - 1]['seq_end'] = $num_read - strlen($line);
|
|
|
|
|
|
// Now that we know where the sequences are in the file we need to add them.
|
|
|
- print "\nStep 2: Importing sequences\n";
|
|
|
+ $this->logMessage("\nStep 2: Importing sequences\n");
|
|
|
for ($i = 0; $i < $num_seqs; $i++) {
|
|
|
$seq = $seqs[$i];
|
|
|
- print "Importing " . ($i + 1) . " of $num_seqs. ";
|
|
|
+ $this->logMessage("Importing " . ($i + 1) . " of $num_seqs. ");
|
|
|
if ($name) {
|
|
|
- print "Current feature: " . $seq['name'] . ".\n";
|
|
|
+ $this->logMessage("Current feature: " . $seq['name'] . ".\n");
|
|
|
}
|
|
|
else {
|
|
|
print "Current feature: " . $seq['uname'] . ".\n";
|
|
|
@@ -572,7 +550,9 @@ class FASTAImporter extends TripalImporter {
|
|
|
|
|
|
loadFastaFeature($fh, $seq['name'], $seq['uname'], $db_id, $seq['accession'], $seq['subject'], $rel_type, $parent_type, $analysis_id, $organism_id, $cvterm, $source, $method, $re_name, $match_type, $parentcvterm, $relcvterm, $seq['seq_start'], $seq['seq_end']);
|
|
|
}
|
|
|
- tripal_set_job_progress($job, 100);
|
|
|
+ if ($this->job) {
|
|
|
+ tripal_set_job_progress($this->job->job_id, 100);
|
|
|
+ }
|
|
|
fclose($fh);
|
|
|
|
|
|
}
|
|
|
@@ -593,8 +573,8 @@ class FASTAImporter extends TripalImporter {
|
|
|
$results = chado_select_record('feature', array('feature_id'
|
|
|
), $values);
|
|
|
if (count($results) > 1) {
|
|
|
- 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));
|
|
|
+ $this->logMessage("Multiple features exist with the name '%name' of type '%type' for the organism. skipping",
|
|
|
+ array('%name' => $name,'%type' => $type), TRIPAL_ERROR);
|
|
|
return 0;
|
|
|
}
|
|
|
if (count($results) == 1) {
|
|
|
@@ -612,8 +592,8 @@ class FASTAImporter extends TripalImporter {
|
|
|
|
|
|
$results = chado_select_record('feature', array('feature_id'), $values);
|
|
|
if (count($results) > 1) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_WARNING, "Multiple features exist with the name '%name' of type '%type' for the organism. skipping", array(
|
|
|
- '%name' => $name,'%type' => $type));
|
|
|
+ $this->logMessage("Multiple features exist with the name '%name' of type '%type' for the organism. skipping",
|
|
|
+ array('%name' => $name,'%type' => $type), TRIPAL_WARNING);
|
|
|
return 0;
|
|
|
}
|
|
|
if (count($results) == 1) {
|
|
|
@@ -622,9 +602,8 @@ class FASTAImporter extends TripalImporter {
|
|
|
|
|
|
// If the feature exists but this is an "insert only" then skip.
|
|
|
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.", array(
|
|
|
- '%name' => $name,'%uname' => $uname,'%type' => drupal_strtolower($match_type)
|
|
|
- ));
|
|
|
+ $this->logMessage("Feature already exists '%name' ('%uname') while matching on %type. Skipping insert.",
|
|
|
+ array('%name' => $name,'%uname' => $uname,'%type' => drupal_strtolower($match_type)), TRIPAL_WARNING);
|
|
|
return 0;
|
|
|
}
|
|
|
}
|
|
|
@@ -649,8 +628,8 @@ class FASTAImporter extends TripalImporter {
|
|
|
);
|
|
|
$success = chado_insert_record('feature', $values);
|
|
|
if (!$success) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to insert feature '%name (%uname)'", array(
|
|
|
- '%name' => $name,'%uname' => $numane));
|
|
|
+ $this->logMessage("Failed to insert feature '%name (%uname)'", array(
|
|
|
+ '%name' => $name,'%uname' => $numane), TRIPAL_ERROR);
|
|
|
return 0;
|
|
|
}
|
|
|
|
|
|
@@ -666,8 +645,8 @@ class FASTAImporter extends TripalImporter {
|
|
|
$feature = $results[0];
|
|
|
}
|
|
|
else {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to retreive newly inserted feature '%name (%uname)'", array(
|
|
|
- '%name' => $name,'%uname' => $numane));
|
|
|
+ $this->logMessage("Failed to retreive newly inserted feature '%name (%uname)'", array(
|
|
|
+ '%name' => $name,'%uname' => $numane), TRIPAL_ERRORR);
|
|
|
return 0;
|
|
|
}
|
|
|
|
|
|
@@ -678,186 +657,175 @@ class FASTAImporter extends TripalImporter {
|
|
|
// if we don't have a feature and the user wants to do an update then fail
|
|
|
if (!$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 " .
|
|
|
- drupal_strtolower($match_type), array('%name' => $name,'%uname' => $uname));
|
|
|
- return 0;
|
|
|
- }
|
|
|
+ $this->logMessage("Failed to find feature '%name' ('%uname') while matching on " . drupal_strtolower($match_type) . ".",
|
|
|
+ array('%name' => $name,'%uname' => $uname), TRIPAL_ERROR);
|
|
|
+ return 0;
|
|
|
+ }
|
|
|
+
|
|
|
+ // if we do have a feature and this is an update then proceed with the update
|
|
|
+ if ($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) {
|
|
|
|
|
|
- // 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
|
|
|
- );
|
|
|
- $results = chado_select_record('feature', array('feature_id'
|
|
|
- ), $values);
|
|
|
+ // 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
|
|
|
+ );
|
|
|
+ $results = chado_select_record('feature', array('feature_id'), $values);
|
|
|
if (count($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.", 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);
|
|
|
- $match = array(
|
|
|
- 'name' => $name,
|
|
|
- 'organism_id' => $organism_id,
|
|
|
- 'type_id' => $cvterm->cvterm_id
|
|
|
- );
|
|
|
-
|
|
|
- // perform the update
|
|
|
- $success = chado_update_record('feature', $match, $values);
|
|
|
- if (!$success) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to update feature '%name' ('%name')", array(
|
|
|
- '%name' => $name,'%uiname' => $uname
|
|
|
- ));
|
|
|
- return 0;
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
+ $this->logMessage("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), TRIPAL_ERROR);
|
|
|
+ return 0;
|
|
|
+ }
|
|
|
+
|
|
|
+ // the changes to the uniquename don't conflict so proceed with the update
|
|
|
+ $values = array('uniquename' => $uname);
|
|
|
+ $match = array(
|
|
|
+ 'name' => $name,
|
|
|
+ 'organism_id' => $organism_id,
|
|
|
+ 'type_id' => $cvterm->cvterm_id
|
|
|
+ );
|
|
|
+
|
|
|
+ // perform the update
|
|
|
+ $success = chado_update_record('feature', $match, $values);
|
|
|
+ if (!$success) {
|
|
|
+ $this->logMessage("Failed to update feature '%name' ('%name')",
|
|
|
+ array('%name' => $name,'%uiname' => $uname), TRIPAL_ERROR);
|
|
|
+ return 0;
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
|
|
|
- // If the user wants to match on the unique name field.
|
|
|
- if (strcmp($match_type, 'Unique name') == 0) {
|
|
|
- // If we're matching on the uniquename and have a new name then
|
|
|
- // we want to update the name.
|
|
|
- $values = array();
|
|
|
- if ($name) {
|
|
|
- $values = array('name' => $name);
|
|
|
- $match = array(
|
|
|
- 'uniquename' => $uname,
|
|
|
- 'organism_id' => $organism_id,
|
|
|
- 'type_id' => $cvterm->cvterm_id
|
|
|
- );
|
|
|
- $success = chado_update_record('feature', $match, $values);
|
|
|
+ // If the user wants to match on the unique name field.
|
|
|
+ if (strcmp($match_type, 'Unique name') == 0) {
|
|
|
+ // If we're matching on the uniquename and have a new name then
|
|
|
+ // we want to update the name.
|
|
|
+ $values = array();
|
|
|
+ if ($name) {
|
|
|
+ $values = array('name' => $name);
|
|
|
+ $match = array(
|
|
|
+ 'uniquename' => $uname,
|
|
|
+ 'organism_id' => $organism_id,
|
|
|
+ 'type_id' => $cvterm->cvterm_id
|
|
|
+ );
|
|
|
+ $success = chado_update_record('feature', $match, $values);
|
|
|
+ if (!$success) {
|
|
|
+ $this->logMessage("Failed to update feature '%name' ('%name')",
|
|
|
+ array('%name' => $name,'%uiname' => $uname), TRIPAL_ERROR);
|
|
|
+ return 0;
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ // Update the residues for this feature
|
|
|
+ loadFastaResidues($fh, $feature->feature_id, $seq_start, $seq_end);
|
|
|
+
|
|
|
+ // add in the analysis link
|
|
|
+ if ($analysis_id) {
|
|
|
+ // if the association doens't alredy exist then add one
|
|
|
+ $values = array(
|
|
|
+ 'analysis_id' => $analysis_id,
|
|
|
+ 'feature_id' => $feature->feature_id
|
|
|
+ );
|
|
|
+ $results = chado_select_record('analysisfeature', array('analysisfeature_id'), $values);
|
|
|
+ if (count($results) == 0) {
|
|
|
+ $success = chado_insert_record('analysisfeature', $values);
|
|
|
if (!$success) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to update feature '%name' ('%name')", array(
|
|
|
- '%name' => $name,'%uiname' => $uname
|
|
|
- ));
|
|
|
+ $this->logMessage("Failed to associate analysis and feature '%name' ('%name')",
|
|
|
+ array('%name' => $name,'%uname' => $uname), TRIPAL_ERROR);
|
|
|
return 0;
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ // now add the database cross reference
|
|
|
+ if ($db_id) {
|
|
|
+ // check to see if this accession reference exists, if not add it
|
|
|
+ $values = array(
|
|
|
+ 'db_id' => $db_id,
|
|
|
+ 'accession' => $accession
|
|
|
+ );
|
|
|
+ $results = chado_select_record('dbxref', array('dbxref_id'), $values);
|
|
|
|
|
|
- // Update the residues for this feature
|
|
|
- loadFastaResidues($fh, $feature->feature_id, $seq_start, $seq_end);
|
|
|
-
|
|
|
- // add in the analysis link
|
|
|
- if ($analysis_id) {
|
|
|
- // if the association doens't alredy exist then add one
|
|
|
- $values = array(
|
|
|
- 'analysis_id' => $analysis_id,
|
|
|
- 'feature_id' => $feature->feature_id
|
|
|
- );
|
|
|
- $results = chado_select_record('analysisfeature', array('analysisfeature_id'), $values);
|
|
|
- if (count($results) == 0) {
|
|
|
- $success = chado_insert_record('analysisfeature', $values);
|
|
|
- if (!$success) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to associate analysis and feature '%name' ('%name')", array(
|
|
|
- '%name' => $name,'%uname' => $uname
|
|
|
- ));
|
|
|
- return 0;
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
-
|
|
|
- // now add the database cross reference
|
|
|
- if ($db_id) {
|
|
|
- // check to see if this accession reference exists, if not add it
|
|
|
- $values = array(
|
|
|
- 'db_id' => $db_id,
|
|
|
- 'accession' => $accession
|
|
|
- );
|
|
|
- $results = chado_select_record('dbxref', array('dbxref_id'), $values);
|
|
|
// if the accession doesn't exist then add it
|
|
|
- if (count($results) == 0) {
|
|
|
- $results = chado_insert_record('dbxref', $values);
|
|
|
- if (!$results) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to add database accession '%accession'", array(
|
|
|
- '%accession' => $accession));
|
|
|
- return 0;
|
|
|
- }
|
|
|
- $results = chado_select_record('dbxref', array('dbxref_id'), $values);
|
|
|
- if (count($results) == 1) {
|
|
|
- $dbxref = $results[0];
|
|
|
- }
|
|
|
- else {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to retreive newly inserted dbxref '%name (%uname)'", array(
|
|
|
- '%name' => $name,'%uname' => $numane));
|
|
|
+ if (count($results) == 0) {
|
|
|
+ $results = chado_insert_record('dbxref', $values);
|
|
|
+ if (!$results) {
|
|
|
+ $this->logMessage("Failed to add database accession '%accession'",
|
|
|
+ array('%accession' => $accession), TRIPAL_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
|
|
|
- );
|
|
|
- $results = chado_select_record('feature_dbxref', array('feature_dbxref_id'), $values);
|
|
|
- if (count($results) == 0) {
|
|
|
- $success = chado_insert_record('feature_dbxref', $values);
|
|
|
- if (!$success) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to add associate database accession '%accession' with feature", array(
|
|
|
- '%accession' => $accession
|
|
|
- ));
|
|
|
+ }
|
|
|
+ $results = chado_select_record('dbxref', array('dbxref_id'), $values);
|
|
|
+ if (count($results) == 1) {
|
|
|
+ $dbxref = $results[0];
|
|
|
+ }
|
|
|
+ else {
|
|
|
+ $this->logMessage("Failed to retreive newly inserted dbxref '%name (%uname)'",
|
|
|
+ array('%name' => $name,'%uname' => $numane), TRIPAL_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
|
|
|
+ );
|
|
|
+ $results = chado_select_record('feature_dbxref', array('feature_dbxref_id'), $values);
|
|
|
+ if (count($results) == 0) {
|
|
|
+ $success = chado_insert_record('feature_dbxref', $values);
|
|
|
+ if (!$success) {
|
|
|
+ $this->logMessage("Failed to add associate database accession '%accession' with feature",
|
|
|
+ array('%accession' => $accession), TRIPAL_ERROR);
|
|
|
return 0;
|
|
|
- }
|
|
|
- }
|
|
|
- }
|
|
|
-
|
|
|
- // now add in the relationship if one exists. If not, then add it
|
|
|
- if ($rel_type) {
|
|
|
- $values = array('organism_id' => $organism_id,'uniquename' => $parent,
|
|
|
- 'type_id' => $parentcvterm->cvterm_id
|
|
|
- );
|
|
|
- $results = chado_select_record('feature', array('feature_id'
|
|
|
- ), $values);
|
|
|
- if (count($results) != 1) {
|
|
|
- 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
|
|
|
- );
|
|
|
- $results = chado_select_record('feature_relationship', array('feature_relationship_id'), $values);
|
|
|
- if (count($results) == 0) {
|
|
|
- $success = chado_insert_record('feature_relationship', $values);
|
|
|
- if (!$success) {
|
|
|
- tripal_report_error('T_fasta_loader', TRIPAL_ERROR, "Failed to add associate database accession '%accession' with feature", array(
|
|
|
- '%accession' => $accession
|
|
|
- ));
|
|
|
- return 0;
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ // now add in the relationship if one exists. If not, then add it
|
|
|
+ if ($rel_type) {
|
|
|
+ $values = array('organism_id' => $organism_id,'uniquename' => $parent, 'type_id' => $parentcvterm->cvterm_id);
|
|
|
+ $results = chado_select_record('feature', array('feature_id'), $values);
|
|
|
+ if (count($results) != 1) {
|
|
|
+ $this->logMessage("Cannot find a unique fature for the parent '%parent' of type '%type' for the feature.",
|
|
|
+ array('%parent' => $parent,'%type' => $parent_type), TRIPAL_ERROR);
|
|
|
+ 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
|
|
|
+ );
|
|
|
+ $results = chado_select_record('feature_relationship', array('feature_relationship_id'), $values);
|
|
|
+ if (count($results) == 0) {
|
|
|
+ $success = chado_insert_record('feature_relationship', $values);
|
|
|
+ if (!$success) {
|
|
|
+ $this->logMessage("Failed to add associate database accession '%accession' with feature",
|
|
|
+ array('%accession' => $accession), TRIPAL_ERROR);
|
|
|
+ return 0;
|
|
|
+ }
|
|
|
}
|
|
|
}
|
|
|
}
|
|
|
-}
|
|
|
|
|
|
/**
|
|
|
* Adds the residues column to the feature.
|
|
|
@@ -871,66 +839,45 @@ class FASTAImporter extends TripalImporter {
|
|
|
* @param $seq_start
|
|
|
* @param $seq_end
|
|
|
*/
|
|
|
- private function loadFastaResidues($fh, $feature_id, $seq_start, $seq_end) {
|
|
|
-
|
|
|
- // First position the file at the beginning of the sequence
|
|
|
- fseek($fh, $seq_start, SEEK_SET);
|
|
|
- $chunk_size = 100000000;
|
|
|
- $chunk = '';
|
|
|
- $seqlen = ($seq_end - $seq_start) + 1;
|
|
|
-
|
|
|
- // Calculate the interval at which we updated the precent complete.
|
|
|
- $interval = intval($seqlen * 0.01);
|
|
|
- if ($interval < 1) {
|
|
|
- $interval = 1;
|
|
|
- }
|
|
|
- // We don't to repeat the update too often or it slows things down, so
|
|
|
- // if the interval is less than 1000 then bring it up to that.
|
|
|
- if ($interval < 100000) {
|
|
|
- $interval = 100000;
|
|
|
- }
|
|
|
- $chunk_intv_read = 0;
|
|
|
- $intv_read = 0;
|
|
|
- $num_read = 0;
|
|
|
- $total_seq_size = 0;
|
|
|
-
|
|
|
- // First, make sure we don't have a null in the residues
|
|
|
- $sql = "UPDATE {feature} SET residues = '' WHERE feature_id = :feature_id";
|
|
|
- chado_query($sql, array(':feature_id' => $feature_id
|
|
|
- ));
|
|
|
-
|
|
|
- // Read in the lines until we reach the end of the sequence. Once we
|
|
|
- // get a specific bytes read then append the sequence to the one in the
|
|
|
- // database.
|
|
|
- print "Sequence complete: 0%. Memory: " . number_format(memory_get_usage()) . " bytes. \r";
|
|
|
- while ($line = fgets($fh)) {
|
|
|
- $num_read += strlen($line) + 1;
|
|
|
- $chunk_intv_read += strlen($line) + 1;
|
|
|
- $intv_read += strlen($line) + 1;
|
|
|
- $chunk .= trim($line);
|
|
|
-
|
|
|
- // If we've read in enough of the sequence then append it to the database.
|
|
|
- if ($chunk_intv_read >= $chunk_size) {
|
|
|
- $sql = "
|
|
|
- UPDATE {feature}
|
|
|
- SET residues = residues || :chunk
|
|
|
- WHERE feature_id = :feature_id
|
|
|
- ";
|
|
|
- $success = chado_query($sql, array(':feature_id' => $feature_id,':chunk' => $chunk
|
|
|
- ));
|
|
|
- if (!$success) {
|
|
|
- return FALSE;
|
|
|
- }
|
|
|
- $total_seq_size += strlen($chunk);
|
|
|
+ private function loadFastaResidues($fh, $feature_id, $seq_start, $seq_end) {
|
|
|
+
|
|
|
+ // First position the file at the beginning of the sequence
|
|
|
+ fseek($fh, $seq_start, SEEK_SET);
|
|
|
+ $chunk_size = 100000000;
|
|
|
+ $chunk = '';
|
|
|
+ $seqlen = ($seq_end - $seq_start) + 1;
|
|
|
+
|
|
|
+ $num_read = 0;
|
|
|
+ $total_seq_size = 0;
|
|
|
+
|
|
|
+ // First, make sure we don't have a null in the residues
|
|
|
+ $sql = "UPDATE {feature} SET residues = '' WHERE feature_id = :feature_id";
|
|
|
+ chado_query($sql, array(':feature_id' => $feature_id));
|
|
|
+
|
|
|
+ // Read in the lines until we reach the end of the sequence. Once we
|
|
|
+ // get a specific bytes read then append the sequence to the one in the
|
|
|
+ // database.
|
|
|
+ while ($line = fgets($fh)) {
|
|
|
+ $num_read += strlen($line) + 1;
|
|
|
+ $chunk_intv_read += strlen($line) + 1;
|
|
|
+ $this->addItemsHandled($num_read);
|
|
|
+ $chunk .= trim($line);
|
|
|
+
|
|
|
+ // If we've read in enough of the sequence then append it to the database.
|
|
|
+ if ($chunk_intv_read >= $chunk_size) {
|
|
|
+ $sql = "
|
|
|
+ UPDATE {feature}
|
|
|
+ SET residues = residues || :chunk
|
|
|
+ WHERE feature_id = :feature_id
|
|
|
+ ";
|
|
|
+ $success = chado_query($sql, array(':feature_id' => $feature_id,':chunk' => $chunk));
|
|
|
+ if (!$success) {
|
|
|
+ return FALSE;
|
|
|
+ }
|
|
|
+ $total_seq_size += strlen($chunk);
|
|
|
$chunk = '';
|
|
|
$chunk_intv_read = 0;
|
|
|
}
|
|
|
- if ($intv_read >= $interval) {
|
|
|
- $percent = sprintf("%.2f", ($total_seq_size / $seqlen) * 100);
|
|
|
- print "Sequence complete: " . $percent . "%. Memory: " . number_format(memory_get_usage()) .
|
|
|
- " bytes. \r";
|
|
|
- $intv_read = 0;
|
|
|
- }
|
|
|
|
|
|
// If we've reached the ned of the sequence then break out of the loop
|
|
|
if (ftell($fh) == $seq_end) {
|
|
|
@@ -945,8 +892,7 @@ class FASTAImporter extends TripalImporter {
|
|
|
SET residues = residues || :chunk
|
|
|
WHERE feature_id = :feature_id
|
|
|
";
|
|
|
- $success = chado_query($sql, array(':feature_id' => $feature_id,':chunk' => $chunk
|
|
|
- ));
|
|
|
+ $success = chado_query($sql, array(':feature_id' => $feature_id,':chunk' => $chunk));
|
|
|
if (!$success) {
|
|
|
return FALSE;
|
|
|
}
|
|
|
@@ -957,12 +903,7 @@ class FASTAImporter extends TripalImporter {
|
|
|
|
|
|
// Now update the seqlen and md5checksum fields
|
|
|
$sql = "UPDATE {feature} SET seqlen = char_length(residues), md5checksum = md5(residues) WHERE feature_id = :feature_id";
|
|
|
- chado_query($sql, array(':feature_id' => $feature_id
|
|
|
- ));
|
|
|
+ chado_query($sql, array(':feature_id' => $feature_id));
|
|
|
|
|
|
- $percent = sprintf("%.2f", ($num_read / $seqlen) * 100);
|
|
|
- print "Sequence complete: " . $percent . "%. Memory: " . number_format(memory_get_usage()) .
|
|
|
- " bytes. \r";
|
|
|
}
|
|
|
-
|
|
|
}
|
Stephen Ficklin