|
|
@@ -0,0 +1,421 @@
|
|
|
+<?php
|
|
|
+
|
|
|
+class data__sequence_record extends ChadoField {
|
|
|
+
|
|
|
+
|
|
|
+ // --------------------------------------------------------------------------
|
|
|
+ // EDITABLE STATIC CONSTANTS
|
|
|
+ //
|
|
|
+ // The following constants SHOULD be set for each descendent class. They are
|
|
|
+ // used by the static functions to provide information to Drupal about
|
|
|
+ // the field and it's default widget and formatter.
|
|
|
+ // --------------------------------------------------------------------------
|
|
|
+
|
|
|
+ // The default label for this field.
|
|
|
+ public static $default_label = 'Sequences';
|
|
|
+
|
|
|
+ // The default description for this field.
|
|
|
+ public static $description = 'A field for displaying all sequences associated with a feature along with their metadata.';
|
|
|
+
|
|
|
+ // Provide a list of instance specific settings. These can be accessed within
|
|
|
+ // the instanceSettingsForm. When the instanceSettingsForm is submitted
|
|
|
+ // then Drupal will automatically change these settings for the instance.
|
|
|
+ // It is recommended to put settings at the instance level whenever possible.
|
|
|
+ // If you override this variable in a child class be sure to replicate the
|
|
|
+ // term_name, term_vocab, term_accession and term_fixed keys as these are
|
|
|
+ // required for all TripalFields.
|
|
|
+ public static $default_instance_settings = [
|
|
|
+ // The short name for the vocabulary (e.g. schema, SO, GO, PATO, etc.).
|
|
|
+ 'term_vocabulary' => 'data',
|
|
|
+ // The name of the term.
|
|
|
+ 'term_name' => 'sequence_record',
|
|
|
+ // The unique ID (i.e. accession) of the term.
|
|
|
+ 'term_accession' => '0849',
|
|
|
+ // Set to TRUE if the site admin is allowed to change the term
|
|
|
+ // type. This will create form elements when editing the field instance
|
|
|
+ // to allow the site admin to change the term settings above.
|
|
|
+ 'term_fixed' => FALSE,
|
|
|
+ ];
|
|
|
+
|
|
|
+ // Indicates the download formats for this field. The list must be the
|
|
|
+ // name of a child class of the TripalFieldDownloader.
|
|
|
+ public static $download_formatters = [
|
|
|
+ 'TripalTabDownloader',
|
|
|
+ 'TripalCSVDownloader',
|
|
|
+ 'TripalNucFASTADownloader',
|
|
|
+ ];
|
|
|
+
|
|
|
+ // The default widget for this field.
|
|
|
+ public static $default_widget = 'data__sequence_record_widget';
|
|
|
+
|
|
|
+ // The default formatter for this field.
|
|
|
+ public static $default_formatter = 'data__sequence_record_formatter';
|
|
|
+
|
|
|
+
|
|
|
+ /**
|
|
|
+ * @see TripalField::elementInfo()
|
|
|
+ */
|
|
|
+ public function elementInfo() {
|
|
|
+ $field_term = $this->getFieldTermID();
|
|
|
+ return [
|
|
|
+ $field_term => [
|
|
|
+ 'operations' => [],
|
|
|
+ 'sortable' => FALSE,
|
|
|
+ 'searchable' => FALSE,
|
|
|
+ 'type' => 'xs:complex',
|
|
|
+ 'readonly' => TRUE,
|
|
|
+ ],
|
|
|
+ ];
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ * @see TripalField::load()
|
|
|
+ */
|
|
|
+ public function load($entity) {
|
|
|
+ $field_name = $this->field['field_name'];
|
|
|
+ $feature = $entity->chado_record;
|
|
|
+
|
|
|
+ // Add the primary sequence from the Chada feature table, residues column.
|
|
|
+ $feature = chado_expand_var($feature, 'field', 'feature.residues');
|
|
|
+
|
|
|
+ // Always add the primary sequence.
|
|
|
+ $this->addPrimary($entity, $feature, $field_name);
|
|
|
+
|
|
|
+ // If this is an mRNA feature then add the gene parent, full length
|
|
|
+ // mRNA, CDS and protein.
|
|
|
+ if ($feature->type_id->name == 'mRNA') {
|
|
|
+ $this->addGeneParent($entity, $feature, $field_name);
|
|
|
+ $featurelocs = $this->addFLmRNA($entity, $feature, $field_name);
|
|
|
+ if (count($featurelocs) > 0) {
|
|
|
+ $this->addCDS($entity, $feature, $field_name, $featurelocs);
|
|
|
+ $this->addProtein($entity, $feature, $field_name);
|
|
|
+ }
|
|
|
+ }
|
|
|
+ // For all others get the sequence from the reference.
|
|
|
+ else {
|
|
|
+ $this->addGenericReference($entity, $feature, $field_name);
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ *
|
|
|
+ * @param unknown $entity
|
|
|
+ * @param unknown $feature
|
|
|
+ * @param unknown $field_name
|
|
|
+ */
|
|
|
+ private function addGenericReference(&$entity, $feature, $field_name) {
|
|
|
+ $label_term = 'rdfs:label';
|
|
|
+ $type_term = 'rdfs:type';
|
|
|
+ $description_term = 'schema:description';
|
|
|
+ $sequence_term = chado_get_semweb_term('feature', 'residues');
|
|
|
+ $seq_coords_term = 'data:2012';
|
|
|
+ $seq_length_term = chado_get_semweb_term('feature', 'seqlen');
|
|
|
+ $seq_md5sum_term = chado_get_semweb_term('feature', 'md5checksum');
|
|
|
+
|
|
|
+ $options = [
|
|
|
+ 'derive_from_parent' => 1,
|
|
|
+ ];
|
|
|
+ $seqs = chado_get_feature_sequences(['feature_id' => $feature->feature_id], $options);
|
|
|
+ $featurelocs = [];
|
|
|
+ foreach ($seqs as $seq) {
|
|
|
+ $featureloc = $this->getFeatureLoc($seq['featureloc_id']);
|
|
|
+ $coords = $this->getSequenceCoords($featureloc);
|
|
|
+
|
|
|
+ $entity->{$field_name}['und'][]['value'] = [
|
|
|
+ $sequence_term => $seq['residues'],
|
|
|
+ $label_term => ucfirst(preg_replace('/_/', ' ', $feature->type_id->name)) . ' Sequence (' . number_format(strlen($seq['residues'])) . 'bp)',
|
|
|
+ $description_term => 'This sequence was extracted from the reference sequence location at ' . $coords['schema:description'] . '.',
|
|
|
+ $seq_coords_term => $coords,
|
|
|
+ $seq_length_term => strlen($seq['residues']),
|
|
|
+ $seq_md5sum_term => md5($seq['residues']),
|
|
|
+ $type_term => $feature->type_id->name
|
|
|
+ ];
|
|
|
+
|
|
|
+ $featurelocs[] = $featureloc;
|
|
|
+ }
|
|
|
+ return $featurelocs;
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ *
|
|
|
+ * @param unknown $entity
|
|
|
+ * @param unknown $feature
|
|
|
+ * @param unknown $field_name
|
|
|
+ */
|
|
|
+ private function addPrimary(&$entity, $feature, $field_name) {
|
|
|
+
|
|
|
+ $label_term = 'rdfs:label';
|
|
|
+ $type_term = 'rdfs:type';
|
|
|
+ $description_term = 'schema:description';
|
|
|
+ $sequence_term = chado_get_semweb_term('feature', 'residues');
|
|
|
+ $seq_length_term = chado_get_semweb_term('feature', 'seqlen');
|
|
|
+ $seq_md5sum_term = chado_get_semweb_term('feature', 'md5checksum');
|
|
|
+
|
|
|
+ if ($feature->residues) {
|
|
|
+ $entity->{$field_name}['und'][]['value'] = [
|
|
|
+ $label_term => 'Primary ' . preg_replace('/_/', ' ', $feature->type_id->name) . 'Sequence (' . number_format($feature->seqlen) . 'bp)',
|
|
|
+ $description_term => 'This is the primary representative sequence for this feature.',
|
|
|
+ $sequence_term => $feature->residues,
|
|
|
+ $seq_length_term => $feature->seqlen,
|
|
|
+ $seq_md5sum_term => $feature->md5checksum,
|
|
|
+ $type_term => $feature->type_id->name
|
|
|
+ ];
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+
|
|
|
+ /**
|
|
|
+ *
|
|
|
+ * @param unknown $entity
|
|
|
+ * @param unknown $feature
|
|
|
+ * @param unknown $field_name
|
|
|
+ */
|
|
|
+ private function addFLmRNA(&$entity, $feature, $field_name) {
|
|
|
+ $label_term = 'rdfs:label';
|
|
|
+ $type_term = 'rdfs:type';
|
|
|
+ $description_term = 'schema:description';
|
|
|
+ $sequence_term = chado_get_semweb_term('feature', 'residues');
|
|
|
+ $seq_coords_term = 'data:2012';
|
|
|
+ $seq_length_term = chado_get_semweb_term('feature', 'seqlen');
|
|
|
+ $seq_md5sum_term = chado_get_semweb_term('feature', 'md5checksum');
|
|
|
+
|
|
|
+ // Sometimes an mRNA may have only exons, only CDS or both exons and
|
|
|
+ // CDS. We need to know which.
|
|
|
+ $sql = "
|
|
|
+ SELECT DISTINCT CVT.name
|
|
|
+ FROM {feature_relationship} FR
|
|
|
+ INNER JOIN {feature} SF on FR.subject_id = SF.feature_id
|
|
|
+ INNER JOIN {feature} OF on FR.object_id = OF.feature_id
|
|
|
+ INNER JOIN {cvterm} CVT on SF.type_id = CVT.cvterm_id
|
|
|
+ WHERE FR.object_id = :feature_id
|
|
|
+ ";
|
|
|
+ $subtypes = chado_query($sql, [':feature_id' => $feature->feature_id])->fetchCol('name');
|
|
|
+
|
|
|
+ $exon = 'exon';
|
|
|
+ if (!in_array('exon', $subtypes) and in_array('CDS', $subtypes)) {
|
|
|
+ $exon = 'CDS';
|
|
|
+ }
|
|
|
+
|
|
|
+ $options = [
|
|
|
+ 'derive_from_parent' => 1,
|
|
|
+ 'aggregate' => 1,
|
|
|
+ 'is_html' => 0,
|
|
|
+ 'sub_feature_types' => ['three_prime_UTR', $exon, 'five_prime_UTR'],
|
|
|
+ ];
|
|
|
+ $seqs = chado_get_feature_sequences(['feature_id' => $feature->feature_id], $options);
|
|
|
+ $featurelocs = [];
|
|
|
+ foreach ($seqs as $seq) {
|
|
|
+
|
|
|
+ $featureloc = $this->getFeatureLoc($seq['featureloc_id']);
|
|
|
+ $coords = $this->getSequenceCoords($featureloc);
|
|
|
+ $types = preg_replace('/_/', ' ', $seq['types']);
|
|
|
+
|
|
|
+ $entity->{$field_name}['und'][]['value'] = [
|
|
|
+ $sequence_term => $seq['residues'],
|
|
|
+ $label_term => 'Full Length mRNA Sequence (' . number_format(strlen($seq['residues'])) . 'bp)',
|
|
|
+ $description_term => 'This full length mRNA sequence was extracted from the reference sequence location at ' . $coords['schema:description'] . ' and contains: ' . implode(', ', $types) ,
|
|
|
+ $seq_coords_term => $coords,
|
|
|
+ $seq_length_term => strlen($seq['residues']),
|
|
|
+ $seq_md5sum_term => md5($seq['residues']),
|
|
|
+ $type_term => 'mRNA'
|
|
|
+ ];
|
|
|
+
|
|
|
+ $featurelocs[] = $featureloc;
|
|
|
+ }
|
|
|
+ return $featurelocs;
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ *
|
|
|
+ * @param unknown $featureloc_id
|
|
|
+ * @return unknown
|
|
|
+ */
|
|
|
+ private function getFeatureLoc($featureloc_id) {
|
|
|
+ $featurelocs_sql = "
|
|
|
+ SELECT SRCF.name, FL.srcfeature_id, FL.strand, FL.fmin, FL.fmax, FL,phase, FL.featureloc_id
|
|
|
+ FROM {featureloc} FL
|
|
|
+ INNER JOIN {feature} SRCF on SRCF.feature_id = FL.srcfeature_id
|
|
|
+ WHERE FL.featureloc_id = :featureloc_id
|
|
|
+ ";
|
|
|
+ return chado_query($featurelocs_sql, [':featureloc_id' => $featureloc_id])->fetchObject();
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ *
|
|
|
+ * @param unknown $featureloc
|
|
|
+ * @return string[]|number[]|NULL[]|unknown[]
|
|
|
+ */
|
|
|
+ private function getSequenceCoords($featureloc) {
|
|
|
+ $description_term = 'schema:description';
|
|
|
+ $reference_term = 'data:3002';
|
|
|
+ $fmin_term = chado_get_semweb_term('featureloc', 'fmin');
|
|
|
+ $fmax_term = chado_get_semweb_term('featureloc', 'fmax');
|
|
|
+ $strand_term = chado_get_semweb_term('featureloc', 'strand');
|
|
|
+
|
|
|
+ $srcfeature = $featureloc->name;
|
|
|
+ $strand = '';
|
|
|
+ if ($featureloc->strand == 1) {
|
|
|
+ $strand = '+';
|
|
|
+ }
|
|
|
+ elseif ($featureloc->strand == -1) {
|
|
|
+ $strand = '-';
|
|
|
+ }
|
|
|
+ else {
|
|
|
+ $strand = '';
|
|
|
+ }
|
|
|
+ $fmin = $featureloc->fmin + 1;
|
|
|
+ $fmax = $featureloc->fmax;
|
|
|
+ $location = $srcfeature . ':' . $fmin . '-' . $fmax . $strand;
|
|
|
+
|
|
|
+ return [
|
|
|
+ $description_term => $location,
|
|
|
+ $reference_term => $srcfeature,
|
|
|
+ $fmin_term => $fmin,
|
|
|
+ $fmax_term => $fmax,
|
|
|
+ $strand_term => $strand,
|
|
|
+ ];
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ *
|
|
|
+ * @param unknown $entity
|
|
|
+ * @param unknown $feature
|
|
|
+ * @param unknown $field_name
|
|
|
+ */
|
|
|
+ private function addCDS(&$entity, $feature, $field_name, $featurelocs) {
|
|
|
+ $label_term = 'rdfs:label';
|
|
|
+ $type_term = 'rdfs:type';
|
|
|
+ $description_term = 'schema:description';
|
|
|
+ $sequence_term = chado_get_semweb_term('feature', 'residues');
|
|
|
+ $seq_length_term = chado_get_semweb_term('feature', 'seqlen');
|
|
|
+ $seq_md5sum_term = chado_get_semweb_term('feature', 'md5checksum');
|
|
|
+
|
|
|
+ foreach ($featurelocs as $featureloc) {
|
|
|
+ $cds_feature = [
|
|
|
+ 'feature_id' => $feature->feature_id,
|
|
|
+ 'parent_id' => $featureloc->srcfeature_id,
|
|
|
+ 'name' => $feature->name,
|
|
|
+ 'featureloc_id' => $featureloc->featureloc_id,
|
|
|
+ ];
|
|
|
+ $options = [
|
|
|
+ 'derive_from_parent' => 1,
|
|
|
+ 'aggregate' => 1,
|
|
|
+ 'sub_feature_types' => ['CDS'],
|
|
|
+ 'is_html' => 0,
|
|
|
+ ];
|
|
|
+
|
|
|
+ $cds_sequence = chado_get_feature_sequences($cds_feature, $options);
|
|
|
+ $coords = $this->getSequenceCoords($featureloc);
|
|
|
+
|
|
|
+ if (count($cds_sequence) > 0) {
|
|
|
+ $entity->{$field_name}['und'][]['value'] = [
|
|
|
+ $label_term => 'Coding Sequence (' . number_format($cds_sequence[0]['length']) . 'bp)',
|
|
|
+ $sequence_term => $cds_sequence[0]['residues'],
|
|
|
+ $description_term => 'This CDS was extracted from the reference sequence location at ' . $coords['schema:description'] . '.' ,
|
|
|
+ $seq_length_term => $cds_sequence[0]['length'],
|
|
|
+ $seq_md5sum_term => md5($cds_sequence[0]['residues']),
|
|
|
+ $type_term => 'CDS'
|
|
|
+ ];
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ *
|
|
|
+ * @param unknown $entity
|
|
|
+ * @param unknown $feature
|
|
|
+ * @param unknown $field_name
|
|
|
+ */
|
|
|
+ private function addGeneParent(&$entity, $feature, $field_name) {
|
|
|
+ $label_term = 'rdfs:label';
|
|
|
+ $type_term = 'rdfs:type';
|
|
|
+ $seq_coords_term = 'data:2012';
|
|
|
+ $description_term = 'schema:description';
|
|
|
+ $sequence_term = chado_get_semweb_term('feature', 'residues');
|
|
|
+ $seq_length_term = chado_get_semweb_term('feature', 'seqlen');
|
|
|
+ $seq_md5sum_term = chado_get_semweb_term('feature', 'md5checksum');
|
|
|
+
|
|
|
+ $sql = "
|
|
|
+ SELECT FO.*
|
|
|
+ FROM {feature_relationship} FREL
|
|
|
+ INNER JOIN {feature} FO on FO.feature_id = FREL.object_id
|
|
|
+ INNER JOIN {cvterm} CVT on CVT.cvterm_id = FO.type_id
|
|
|
+ INNER JOIN {cvterm} RCVT on RCVT.cvterm_id = FREL.type_id
|
|
|
+ WHERE
|
|
|
+ FREL.subject_id = :feature_id and
|
|
|
+ CVT.name = 'gene' and
|
|
|
+ RCVT.name IN ('part_of')
|
|
|
+ ";
|
|
|
+ $genes = chado_query($sql, [':feature_id' => $feature->feature_id]);
|
|
|
+ while ($gene = $genes->fetchObject()) {
|
|
|
+ if (!empty($gene->residues)) {
|
|
|
+ $entity->{$field_name}['und'][]['value'] = [
|
|
|
+ $label_term => 'Gene Sequence (primary)',
|
|
|
+ $sequence_term => $gene->residues,
|
|
|
+ $label_term => 'The gene sequence.',
|
|
|
+ $seq_length_term => strlen($gene->residues),
|
|
|
+ $seq_md5sum_term => md5($gene->residues),
|
|
|
+ $type_term => 'polypeptide'
|
|
|
+ ];
|
|
|
+ }
|
|
|
+ else {
|
|
|
+ $seqs = chado_get_feature_sequences(['feature_id' => $gene->feature_id], ['derive_from_parent' => 1]);
|
|
|
+ foreach ($seqs as $seq) {
|
|
|
+ $featureloc = $this->getFeatureLoc($seq['featureloc_id']);
|
|
|
+ $coords = $this->getSequenceCoords($featureloc);
|
|
|
+ $entity->{$field_name}['und'][]['value'] = [
|
|
|
+ $label_term => 'Gene Sequence (' . number_format(strlen($seq['residues'])) . 'bp)',
|
|
|
+ $sequence_term => $seq['residues'],
|
|
|
+ $description_term => 'This gene sequence was extracted from the reference sequence location at ' . $coords['schema:description'] . '.' ,
|
|
|
+ $seq_coords_term => $coords,
|
|
|
+ $seq_length_term => strlen($seq['residues']),
|
|
|
+ $seq_md5sum_term => md5($seq['residues']),
|
|
|
+ $type_term => 'gene'
|
|
|
+ ];
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ /**
|
|
|
+ *
|
|
|
+ * @param unknown $entity
|
|
|
+ * @param unknown $feature
|
|
|
+ * @param unknown $field_name
|
|
|
+ */
|
|
|
+ private function addProtein(&$entity, $feature, $field_name) {
|
|
|
+ $label_term = 'rdfs:label';
|
|
|
+ $type_term = 'rdfs:type';
|
|
|
+ $description_term = 'schema:description';
|
|
|
+ $sequence_term = chado_get_semweb_term('feature', 'residues');
|
|
|
+ $seq_length_term = chado_get_semweb_term('feature', 'seqlen');
|
|
|
+ $seq_md5sum_term = chado_get_semweb_term('feature', 'md5checksum');
|
|
|
+
|
|
|
+ $sql = "
|
|
|
+ SELECT F.*
|
|
|
+ FROM {feature_relationship} FR
|
|
|
+ INNER JOIN {feature} F on FR.subject_id = F.feature_id
|
|
|
+ INNER JOIN {cvterm} CVT on CVT.cvterm_id = F.type_id
|
|
|
+ INNER JOIN {cvterm} RCVT on RCVT.cvterm_id = FR.type_id
|
|
|
+ WHERE
|
|
|
+ FR.object_id = :feature_id and
|
|
|
+ CVT.name = 'polypeptide' and
|
|
|
+ RCVT.name IN ('derives_from', 'part_of')
|
|
|
+ ORDER BY FR.rank ASC
|
|
|
+ ";
|
|
|
+ $proteins = chado_query($sql, [':feature_id' => $feature->feature_id]);
|
|
|
+ while ($protein = $proteins->fetchObject()) {
|
|
|
+ if (!empty($protein->residues)) {
|
|
|
+ $entity->{$field_name}['und'][]['value'] = [
|
|
|
+ $label_term => 'Protein Sequence (' . number_format(strlen($protein->residues)) . 'aa)',
|
|
|
+ $sequence_term => $protein->residues,
|
|
|
+ $description_term => 'The protein sequence.',
|
|
|
+ $seq_length_term => strlen($protein->residues),
|
|
|
+ $seq_md5sum_term => md5($protein->residues),
|
|
|
+ $type_term => 'polypeptide'
|
|
|
+ ];
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+}
|
|
|
+
|
Stephen Ficklin